ADVANCED PROGRAM | ICOICT 2021
1
CONFERENCEINFORMATIONDates August,3–52021
Organizer TelkomUniversity,Indonesia
Co-organizer MultimediaUniversity,Malaysia
UniversitasGadjahMada,Indonesia
Venue ZoomVirtualWebinarandMeeting
Secretariat SchoolofComputing,TelkomUniversity
PanambulaiBuilding
Jl.TelekomunikasiTerusanBuahBatu,Bandung
WestJava40257
Indonesia
Phone:+62227569131
Fax:+62227565930
Email:[email protected]
ConferenceWebsite www.icoict.org
ADVANCED PROGRAM | ICOICT 2021
2
WELCOMEMESSAGEOn behalf of the Organizing and Program Committee, we warmly welcome you to the 9thInternational Conference on Information and Communication Technology (ICoICT) 2021 onAugust3-5th,2021.The9th ICoICT2021 is jointlyorganizedbyTelkomUniversity Indonesia,MultimediaUniversityMalaysia,andUniversitasGadjahMadaIndonesia,inassociationwithIEEEIndonesiaSection,IEEESignalProcessingSocietyChapter,andTheIndonesiaSectionComputerSocietyChapter.ThepreviousICoICTconferenceshavesuccessfullyservedasaforumtobringtogetheradiversegroupofpeoplefromacademicsandindustriestoshareandpresentthelatestissuesandrecentdevelopments in InformationandCommunicationTechnology(ICT).PapersfromthepreviousICoICT2013until2020havebeenpublishedinIEEEXploreandindexedinScopus.Thetechnicalprogramof9thICoICT2021consistsofeightkeynotes,oneknowledgetransfer,fivetutorials,sixtrackson“DigitalInnovationsforPost-pandemicRecovery”,and25parallelsessions.Forthefirsttime,theconferencefeaturessocialeventsthatallowconferenceparticipantstomeetanddiscusswithfellowresearchersinthesamefieldofinterest.CompetitionfortheBestPaperAwardisalsoorganized.The9thICoICT2021received296papersubmissionsfrom20countries,outofwhich122papershavebeenaccepted–correspondingtoanacceptancerateof43.4%.Allpapersubmissionshavebeensubjectedtoarigorouspeer-reviewprocessthatevaluatestheirsignificance,novelty,andtechnicalquality.Eachpaperwasreviewedindependentlybyatleastthreeexperts.Due to the COVID-19 pandemic, we have decided to hold the 9th ICoICT 2021 as a virtualconference.Theorganizingcommitteehadbeenworkhardtocreateavirtualconferencethatwillbevaluableandengagingforbothpresentersandattendees.Thefullconferenceformatmixespre-recorded and asynchronous engagement and lives engagement through Question-and-Answer(Q&A).The9thICoICT2021hasbeenorganizedasaresultoftheworkandeffortofcolleagues,friends,andorganizations.Wewishtothankallwhohaveparticipatedandsupportedourworkinmanywaysandallwhohelpedusmakethiseventpossibleandsuccessful.WewouldliketoexpressourgratitudetotheOrganizingCommitteeandTechnicalCommitteemembersandallTelkomUniversitycolleagueswhoassistedusinplanningandorganizingthisconference.Wealsowishtothankallthereviewerswhoworkedveryhardinreviewingpapersandprovidingsuggestionsforthepaper’simprovements.WewouldliketoexpressoursinceregratitudetotheKeynoteandTutorialSpeakers.Wewouldalsoliketothankallofthesponsoringorganizationsforprovidingtheirgenerous financial support.Lastbutnot least,wewould like togiveappreciation to theauthorswhohavesubmittedtheirexcellentworkstothisconferenceandalltheattendees.Weappreciateyourvirtualattendanceat the9th ICoICT2021.Wehopeyouenjoyall thekeynotesessions,thetechnicalsessions,andthesocialeventsandinspireyourfutureresearch.Dr.WarihMaharaniGeneralChairofICoICT2021
ADVANCED PROGRAM | ICOICT 2021
3
COMMITTEESteeringCommittee
§ AAdiwijaya,TelkomUniversity,Indonesia§ AhmadRafi,MultimediaUniversity,Malaysia§ AfizanAzman,MelakaInternationalCollegeofScienceandTechnologyMalaysia§ AriMoesriamiBarmawi,TelkomUniversity,Indonesia§ HairulA.Abdul-Rashid,MultimediaUniversity,Malaysia§ SongHoeLau,MultimediaUniversity,Malaysia§ MamanAbdurohman,TelkomUniversity,Indonesia§ ParmanSukarno,TelkomUniversity,Indonesia§ RinaPudjiastuti,TelkomUniversity,Indonesia§ ShafinarIsmail,UniversitiTeknologiMara,Malaysia§ SyedAbdulRahmanAlHaddad,UniversitiPutraMalaysia,Malaysia§ KikiMaulanaAdhinugraha,LaTrobeUniversity,Australia§ SultanAlamri,SaudiElectronicUniversity,SaudiArabia
OrganizingCommitteeGeneralChair
§ WarihMaharani,TelkomUniversity,IndonesiaGeneralCo-Chair
§ VeraSuryani,TelkomUniversity,Indonesia§ OngThianSong,MultimediaUniversity,Malaysia
SecretariatChair
§ SitiKarimah,TelkomUniversity,Indonesia§ OoiShihYin,MultimediaUniversity,Malaysia§ DitaOktaria,TelkomUniversity,Indonesia
PublicationChair
§ DawamDwiJatmiko,TelkomUniversity,Indonesia§ AndityaArifianto,TelkomUniversity,Indonesia§ SitiZainabIbrahim,MultimediaUniversity,Malaysia§ Mohd.FikriAzlibinAbdullah,MultimediaUniversity,Malaysia
FinanceChair
§ AnnisaAditsania,TelkomUniversity,Indonesia§ ChongSiewChin,MultimediaUniversity,Malaysia§ SitiSa’adah,TelkomUniversity,Indonesia§ RamantiDharayani,TelkomUniversity,Indonesia
EventandLogisticChair
§ FazmahArif,TelkomUniversity,Indonesia§ MiraSabariah,TelkomUniversity,Indonesia
ADVANCED PROGRAM | ICOICT 2021
4
§ DonniRichasdy,TelkomUniversity,Indonesia§ AndiDarmawan,UniversitasGadjahMada,Indonesia
CallForPaperChair
§ AgungTotoWibowo,TelkomUniversity,Indonesia§ AuliaKhamasHeikhmakhtiar,TelkomUniversity,Indonesia§ MahmudDwiSulistiyo,TelkomUniversity,Indonesia
TutorialandSpecialSessionChair
§ PutuHarryGunawan,TelkomUniversity,IndonesiaSponsorshipChair
§ KemasRahmatSalehWihaja,TelkomUniversity,Indonesia§ RizkaRezaPahlevi,TelkomUniversity,Indonesia
Webmaster
§ RahmatYasirandi,TelkomUniversity,Indonesia§ MuhammadAlMakky,TelkomUniversity,Indonesia
TechnicalProgramCommitteeTPCChair
§ AdeRomadhony,TelkomUniversity,IndonesiaTrackChair
§ DiditAdytia,TelkomUniversity,Indonesia§ TeeConnie,MultimediaUniversity,Malaysia§ KusumaAyuLaksitowening,TelkomUniversity,Indonesia§ PangYingHan,MultimediaUniversity,Malaysia§ MardhaniRiasetiawan,UniversitasGadjahMada,Indonesia§ DodyQoriUtama,TelkomUniversity,Indonesia
ReviewersAchmadRizal,TelkomUniversityAdeCandra,UniversityofSumateraUtaraAdeRomadhony,TelkomUniversityAdityaFirmanIhsan,TelkomUniversityAfridaHelen,UniversitasPadjadjaran(Unpad)AgungDewandaru,BandungInstituteofTechnologyAgungWibowo,TelkomUniversityAgusHartoyo,TelkomUniversityAlbertusJokoSantoso,UniversitasAtmaJayaYogyakartaAlfanWicaksono,UniversitasIndonesiaAlfianAkbarGozali,TelkomUniversityAmarilisPutriYanuarifiani,MultimediaUniversityAndiWahjuRahardjoEmanuel,UniversitasAtmaJayaYogyakartaAndryAlamsyah,TelkomUniversityAngelinaKurniati,TelkomUniversity
ADVANCED PROGRAM | ICOICT 2021
5
ApriSiswanto,UniversitasIslamRiauArbiNasution,UniversitasIslamRiauAriBarmawi,TelkomUniversityArieArdiyanti,TelkomUniversityArnidaLatifah,IndonesianInstituteofScienceAsepWahyudin,UniversitasPendidikanIndonesia(UPIAshokKumarVeerasamy,UniversityofTurkuBayuErfianto,TelkomUniversityBedyPurnama,TelkomUniversityBerliyanto,InstitutTeknologiBudiUtomoBonyParulianJosaphat,SekolahTinggiIlmuStatistikBudiPrasetya,TelkomUniversityCatherineSereati,UniversitasKatolikIndonesiaAtmaJayaCeciliaEstiNugraheni,ParahyanganCatholicUniversityChinPooLee,MultimediaUniversityChooKimTan,MultimediaUniversityChristianNaa,ParahyanganCatholicUniversityChristinaJuliane,STMIKAMIKBANDUNGDadeNurjanah,TelkomUniversityDanaSulistyoKusumo,TelkomUniversityDanySaputra,UniversitasBinaNusantaraDedyRahmanWijaya,TelkomUniversityDeniSaepudin,TelkomUniversityDerwinSuhartono,BinaNusantaraUniversityDhomasHattaFudholi,UniversitasIslamIndonesiaDianaPurwitasari,InstitutTeknologiSepuluhNopemberDidiRosiyadi,IndonesianInstituteofSciencesDiditAdytia,TelkomUniversityDodyUtama,TelkomUniversityDumaHutapea,UniversitasKatolikIndonesiaAtmaJayaJakartaDwizaRiana,STMIKNusaMandiriJakartaEmaRachmawati,TelkomUniversityEmilKaburuan,BinaNusantaraUniversityEndahSudarmilah,UniversitasMuhammadiyahSurakartaEndangChumaidiyah,TelkomUniversityEndangDjuana,TrisaktiUniversityErnaNababan,UniversityofNorthSumateraErwinB.Setiawan,TelkomUniversityEsaPrakasa,IndonesianInstituteofSciencesEvizalAbdulKadir,UniversitasIslamRiauFakhriyHario,UniversitasBrawijayaFang-FangChua,MultimediaUniversityFarahAfianti,TelkomUniversityFarizDarari,UniversitasIndonesiaFavianDewanta,TelkomUniversityFazmahArifYulianto,TelkomUniversityFettyLubis,BandungInstituteofTechnologyFordGaol,BinaNusantaraUniversityGandevaBayuSatrya,TelkomUniversityGentaIndraWinata,HongKongUniversityofScienceandTechnologyGoHasegawa,TohokuUniversityGohKahOngMichael,MultimediaUniversity
ADVANCED PROGRAM | ICOICT 2021
6
HapnesToba,MaranathaChristianUniversityHarrySantoso,UniversitasIndonesiaHengSiongLim,MultimediaUniversityHilalNuha,TelkomUniversityIGustiBagusBaskaraNugraha,BandungInstituteofTechnologyImeldaAtastina,TelkomUniversityIndraBudi,UniversitasIndonesiaIndraChandra,TelkomUniversityInnaSyafarina,IndonesianInstituteofScienceInneHusein,TelkomUniversityIntanNurmaYulita,UniversitasPadjadjaranIrwanIrwan,STMIKMDPIsmanKurniawan,TelkomUniversityIzzatdinAbdulAziz,UniversitiTeknologiPETRONASJanWiraGotamaPutra,TokyoInstituteofTechnologyJessicaPermatasari,MultimediaUniversityJi-JianChin,MultimediaUniversityJimmyTirtawangsa,TelkomUniversityKarelBachri,AtmaJayaCatholicUniversityofIndonesiaKemasMuslimLhaksmana,TelkomUniversityKemasRahmatSalehWiharja,TelkomUniversityKianMingLim,MultimediaUniversityKikiMaulanaAdhinugraha,LaTrobeUniversityKorediantoUsman,TelkomUniversityKurnianingsihKurnianingsih,PoliteknikNegeriSemarangKusumaLaksitowening,TelkomUniversityLeeYingChong,MultimediaUniversityLee-YengOng,MultimediaUniversityLeonardGoeirmanto,MercuBuanaUniversityLeowMengChew,MultimediaUniversityLillianYeeKiawWang,MonashUniversityMalaysiaMahmudSulistiyo,TelkomUniversityMamanAbdurohman,TelkomUniversityMarisaParyasto,TelkomUniversityMasayuLeyliaKhodra,InstitutTeknologiBandungMewatiAyub,MaranathaChristianUniversityMickeRusmerryani,ResearchLabAPChannelMiraKaniaSabariah,TelkomUniversityMochArifBijaksana,TelkomUniversityMohammedS.Al-Abiad,UniversityofBritishColumbiaMuhamadKoyimatu,UniversitasPertaminaMuhamadRisqiUtamaSaputra,UniversityofOxfordMuhammadAgniCatur,BhaktiSampoernaUniversityMuhammadHaris,BukalapakMuhammadJohanAlibasa,TelkomUniversityMuhammadKusban,UniversitasGadjahMadaMuhammadNurKholishAbdurrazaq,InstitutAgamaIslamAz-ZaytunMustafaMatDeris,UniversityTunHusseinOnnMalaysiaNianChiTay,MultimediaUniversityNikenDwiWahyuCahyani,TelkomUniversityNiruwanTurnbull,MahasarakhamUniversityNoorCholisBasjaruddin,PoliteknikNegeriBandung
ADVANCED PROGRAM | ICOICT 2021
7
NungkiSelviandro,TelkomUniversityNurulAin,UniversitiTeknologiMaraOngThianSong,MultimediaUniversityOpimSitompul,UniversityofNorthSumateraPancaPutra,UniversitasIndonesiaParmanSukarno,TelkomUniversityPaulusInsapSantosa,UniversitasGadjahMadaPeyYunGoh,MultimediaUniversityPutuHarryGunawan,TelkomUniversityRadoslavaKraleva,South-WestUniversityRayAdderleyJMGining,UniversitiTeknologiMARARezaPulungan,UniversitasGadjahMadaRimbaWhidianaCiptasari,TelkomUniversityRinaMardiati,UINSunanGunungJatiRioGunturUtomo,TelkomUniversityRisnandar,IndonesianInstituteofSciencesRobertusNugroho,SoegijapranataCatholicUniversityRonaldAdrian,UniversitasGadjahMadaRonyKalfarisi,BentleySystemsRuhailaMaskat,UniversitiTeknologiMARASandyVantika,BandungInstituteofTechnologySaparudinSaparudin,TelkomUniversitySariDewiBudiwati,TelkomUniversitySatriaMandala,TelkomUniversitySeptafiansyahDwiPutra,PoliteknikNegeriLampungSetiawanHadi,UniversitasPadjadjaranSetyorini,TelkomUniversityShenYeePang,MultimediaUniversityShihYinOoi,MultimediaUniversityShingChyiChua,MultimediaUniversityShukorSanimMohdFauzi,UniversitiTeknologiMara,PerlisCampusSiewChinChong,MultimediaUniversitySinYinTan,MultimediaUniversitySiongHoeLau,MultimediaUniversitySitiFatimahAbdulRazak,MultimediaUniversitySitiZainabIbrahim,MultimediaUniversitySritrustaSukaridhoto,PoliteknikElektronikaNegeriSurabayaSusetyoBagasBhaskoro,PoliteknikManufakturBandungSuyanto,TelkomUniversitySwee-HuayHeng,MultimediaUniversityTajulRosliRazak,UniversitiTeknologiMARATantyOktavia,BinaNusantaraUniversityTedjoDarmanto,STMIKAMIKBandungTeeConnie,MultimediaUniversityThomasBasuki,CurtinUniversityThomhertSiadari,DDHInc.TimilehinAderinola,MultimediaUniversityTiranaNoorFatyanosa,BrawijayaUniversityTohariAhmad,InstitutTeknologiSepuluhNopemberToraFahrudin,TelkomUniversityVeraSuryani,TelkomUniversityWarihMaharani,TelkomUniversity
ADVANCED PROGRAM | ICOICT 2021
8
WarsunNajib,UniversitasGadjahMadaWeeHowKhoh,MultimediaUniversityWeiChuenYauXiamen,UniversityMalaysiaWikkyAlMaki,TelkomUniversityWisnuAnantaKusuma,BogorAgricultureInstituteWiwinSuwarningsih,IndonesianInstituteofScienceYantiRusmawati,TelkomUniversityYeePingLiew,MultimediaUniversityYeeYenYuen,MultimediaUniversityYingHanPang,MultimediaUniversityYudhaPurwanto,TelkomUniversityYudiWibisono,IndonesiaUniversityofEducationYuliantSibaroni,TelkomUniversityYunitaSari,UniversitasGadjahMadaZ.k.AbdurahmanBaizal,TelkomUniversityZoyaPourmirza,NewcastleUniversity
ADVANCED PROGRAM | ICOICT 2021
9
VIRTUALCONFERENCEGENERALINSTRUCTIONS
GeneralInformationICoICT2021will be hosted as a fully virtual conferenceusing Zoom.There are four types ofsessionsinICoICT2021,namely:keynotespeeches,tutorials,Birds-of-a-feather(BoF)sessions,andpaperpresentationsessions.Eachsessionwillhaveaco-host,amoderatororsessionchairorsessionhost,presenters,andattendees.Theco-hostwillbethepersonthatfacilitatestheuseof the technology. The moderator or session chair is present to coordinate the session andmanageQ&A.Accesstosessions• Keynotespeechesandtutorialswillbeheld in theZoomWebinarroomwhichhasa large
enoughcapacityandwillalsobelivestreamedviaYoutubesothatitcanbefreelyaccessedbythepublic.
• Birds-of-a-feathersessionsandparallelpaperpresentationswillbeheldintheZoomMeetingroomswhicharesmallerincapacityandcanonlybeattendedbypresentersandauthorsofregisteredpapers.
BasicSessionStructure• 10minutesbeforetheSession:Theco-hostwillstartthedesignatedZoomsession.• 2minutesbeforetheSession:Theco-hostwillstarttherecording.• 1minutebeforetheSession:Themoderatororsessionchairintroducesthesession.• BeginningoftheSession:Themoderatororsessionchairwillintroducethepresenter.
ThetalkwillbepresentedusingthevideorecordingsubmittedbythepresenterfollowedbyliveQ&A.Thedurationofeachpresentationbytypearelistedbelow:
o Keynotespeech:(1hour)Recordedpresentation(video)=40minutesLiveQ&A=10minutes
o Tutorial:(1.5hours)Recordedvideo=40–60minutesLiveQ&A=20-40minutes
o Birds-of-a-feather:(2–2.5hours)o Paperpresentation:(20minutes)
Recordedpresentation(video)=15minutesLiveQ&A=5minutes
• EndoftheSession:Themoderatororsessionchairwillconcludethesessionandcheckattendance,thentheco-hostwillendthesession.
ScheduleandTimeAlltimeintheprogramscheduleisinWesternIndonesiaTime/WaktuIndonesiaBarat(WIB;UTC+7).Pleasemakeadjustmentstothetimestatedinthescheduleintoyourlocaltime.CurrenttimeinWIBcanbefoundonhttp://time.bmkg.go.id/.
ADVANCED PROGRAM | ICOICT 2021
10
CONFERENCESCHEDULE
08.15–08.30 SoftOpeningbyDeanofSchoolofComputing,TelkomUniversity
(Dr.Z.K.AbdurahmanBaizal)
08.30–09.30 KeynoteSpeech1:Prof.HaryadiS.Gunawi(UniversityofChicago)WhyDoestheCloudStopComputing?LessonsfromThousandsofBugReports,ServiceOutages,andAnecdotalEvidence
09.30–11.30 Tutorial1:JuneidiTsai(MicrosoftIndonesia)MicrosoftEducationTransformationFrameworkforHigherEducation
11.30–12.30 Break
12.30–13.30 KeynoteSpeech2:Dr.DavidTaniar(MonashUniversity)ContactTracingduringCovid-19Pandemic:AnAustralianExperience
13.30–14.30 KeynoteSpeech3:SolehAyubi,Ph.D(ChiefDigitalHealthcareOfficeratBioFarmaIndonesia)Covid-19VaccineDistribution:anexampleofAcceleratinghealthcaretransformation
14.30–14.45 Break14.45–17.45 BirdsofaFeatherSessions
RoomYudhistira:Dr.MayaAriyanti
"ELearning"
RoomBima:Afiahayati,Ph.D"AIandBioinformaticsforCOVID-19"
RoomArjuna:KemasRahmatShaleh,Ph.D"KnowledgeManagement,KnowledgeRepresentationandreasoning"
RoomNakula:Dr.AgusHartoyo"ArtificialIntelligenceofThings(AIoT):TheintersectionofAIandIoT"
RoomSadewa:Dr.MuhammadJohanAlibasa“DigitalfootprintsduringCOVID-19pandemics”
Day 1, Tuesday 3 August 2021
ADVANCED PROGRAM | ICOICT 2021
11
CONFERENCESCHEDULE08.00–08.10 OpeningCeremony:Greetingsandvideopresentation08.10–08.20 WelcomeSpeechbyGeneralChairofICoICT2021
Dr.WarihMaharani08.20–08.25 SpeechbyDeanoftheFacultyFacultyofMathematicsandNaturalScience,
UniversitasGadjahMadaProf.Dr.Triyono,S.U.
08.25–08.30 Speech by Dean of the Faculty of Information Science and Technology,MultimediaUniversityProf.Ts.Dr.LauSiongHoe
08.30–08.35 SpeechbyChairofIEEEIndonesiaSectionDr.-IngWahyudiHasbi,S.Si,M.Kom
08.35–08.40 OpeningSpeechbyRectorofTelkomUniversityProf.Dr.Adiwijaya
08.40–08.45 PhotoSessions
08.45–09.45 KeynoteSpeech4:Dr.Ir.HammamRizaM.Sc,IPU(BPPT)DigitalInnovationsforPost-PandemicRecovery
09.45–10.45 KnowledgeTransfer:WuShiwei(HUAWEICLOUD-CTOofAPACRegion)WorkforceEnablementintheNewNormalofPost-Pandemic,PoweredByCloudTechnology
10.45–11.00 Break11.00–13.00 Tutorial2:Dr.BayuErfianto(TelkomUniversity)
Chest-wallmotiontrackingusingInertialMeasurementUnits(IMUs)
Parallelsessions#1
13.00–14.00 Break14.00–16.00 Tutorial3:MohEdiWibowo,S.Kom.,M.Kom.,Ph.D.(UGM)
Faceanalysisinpublicspace
Parallelsessions#2
16.00–17.00 KeynoteSpeech5:Prof.Dr.KazemRahimi(TheUniversityofOxford)Machine learning and digital technologies in the context ofepidemiologyandclinicaltrials
17.00–18.00
Keynote Speech 6: Dr. Mohamad Hardyman Barawi (Universiti MalaysiaSarawak)ModellingsentimentandcontrastiveopinionofCOVID-19PandemiconSocialMedia:LinkingComputerScienceandSocialScience
18.00–18.10 Closingday2
Day 2, Wednesday 4 August 2021
ADVANCED PROGRAM | ICOICT 2021
12
CONFERENCESCHEDULE
08.00–08.15 Openingday308.15–10.15 Tutorial4:DhanySaputra,Ph.D(Brüel&KjærVibro)
Bioinformaticshashelpedprevent&controlapandemic.What'snext?
Parallelsessions#3
10.15–12.15 Tutorial5:Assoc.Prof.Dr.Md.ShohelSayeed(MMU)BigData:Trends,Challenges&Opportunities
Parallelsessions#4
12.15–13.15 Break
13.15–15.15 ParallelSessions#5
15.15–15.30 Break
15.30–16.30 KeynoteSpeech7:Prof.HadiSusanto(UniversityofEssex)COVID-19modellinginIndonesia:Amathematician’sapology
16.30–17.30 KeynoteSpeech8:RezaKhorshidi,D.Phil.((TheUniversityofOxford)
17.30–17.35 Greetings17.35–17.45 BestpaperawardbyICoICT2021TPCChair
Dr.AdeRomadhony
17.45–17.50 ICoICT2022presentation
17.50–17.55 ClosingbyViceDeanofSchoolofComputingTelkomUniversityParmanSukarno,Ph.D
Day 3, Thursday 5 August 2021
ADVANCED PROGRAM | ICOICT 2021
13
INSTRUCTIONFORMODERATORS
ANDSESSIONCHAIRS
• Generalresponsibilityofamoderatororsessionchair:
1. Introduce the session, explain the structure of the session, rule of theQ&A, and then
introduceeachspeakerorpresenter/paper-titlebeforethevideoplayback;
2. TakechatquestionsfromQ&Apanelandchatpanelduringthepresentationplayback;
3. Read the questions to the presenting speaker or author during the Q&A periodwith
absolutelynodelayontheallocatedQ&Aslots;
4. Concludethesessionandcheckattendance;
5. Makesurethetimeisnotviolated
• Organizerwill sendyouan invitation linkviaemail. Justclickon the link to join theZoom
session.
• Pleasejointhesessionatleast10minutesinadvance,pleasetestyourmicrophoneoncejoined
sothatthesessioncanstartontime.
• We recommend that you turn on your video to engage the attendees during the sessional
introduction.
• Afterintroducingthesessionandthepresenter,werecommendyoutoturnoffthevideoso
thattheattendeescanfocusonthetalkvideo.
• Asorganizers,wewouldliketoensureasmoothandproductivevirtualconference.
• Duringthereplayofpresentationvideo,pleasekeeptrackofthequestionsontheQ&Aand
chatpanel.
• Afterthevideoplayback,unmuteyourselfandshareyourwebcam.Thehostwillalsounmutes
andsharescorrespondingspeaker’sorpresenter’swebcam.
• PleasemakesuretoaskorallythequestionsandaccordingtotheFIFOtimetheywerefirst
submitted.Iftherearenotmanyquestions,feelfreetoasksomeofyourown.
• Sometimes the audience may need to clarify their question. In that case, it is upon the
discretion of the Session Chair to unmute the attendeewho placed the question tomake
clarifications.
• PleasebemindfuloftheQ&Atimelimits!Wecannotintroducedelaysonthepredetermined
slotsofeachsession,itwillpushothersessionsbehind.
• Afterthepresenteraddressedaquestion,pleaseindicateitisansweredverbally.
• Thescreenwillcontaincuestohelpyougoverntheflowofthesession.
ADVANCED PROGRAM | ICOICT 2021
14
INSTRUCTIONFORPRESENTERS
• Organizerwillsendyouaninvitationlinkasapresenterviaemail.Justclickonthelinktojoin
theZoomsession.
• Pleasejoinatleast5minutesbeforeyourdesignatedtimeslotandbepresentduringthevideo
playbackaswellastheQ&Asession.
• Asorganizers,wewouldliketoensureasmoothandproductivevirtualconference.Following
thevideopresentationthereisashortQ&Asession.
• Yourmicrophonewillbemutedduringthevideoplayback.
• Duringthereplayofthepresentationvideo,attendeesorotherpresenterswillaskquestions
related to the presentation through theQ&Apanel (for attendee) or chat panel (for other
presenters).
• You are encouraged to keep an eye on the questions so to answer them during the Q&A
Session.
• Attheendofthepre-recordedpresentation,thehostwillunmuteyourmicrophoneandshares
your webcam. The moderator or session chair will then ask you to answer some of the
questionsinsequencetheyweresubmittedandwithintheallottedQ&Aperiod.
• Themoderatororsessionchairwilltrytocoverasmanyquestionsaspossibledependingon
theallottedtime.Incase,someofthequestionshavenotbeenansweredattendeesmaywant
todiscusswiththepresenteroff-line.
• Youarewelcometostayinthesessionasanattendeewhennotpresenting.
• Whenyouarenotpresentingyourpaper,youcanalsoaskquestionstootherpresentersvia
chatpanel.Beginyourtextwith“[ASK]”toindicateitasaquestiontothepresenter.Senditto
sessionchair(orallpanelistifyouprefertodoso).
• Youarealsowelcometoregisterasanattendeeforanyothersession/eventthatinterests
you (including keynote speeches, tutorials, Birds-of-a-feather sessions, and other paper
presentations)usingregistrationlinksprovidedintheicoict.org.Pleasenotethateachroom
haslimitedcapacity.
Breaks
Conferencing, online and in-person can be exhausting!Weneed to take breaks.Wewill take
breaks.Breaksarebuilt-intotheschedule!
• Standupandstretch,getasnack,comebackrefreshed!
• IfyouleaveyourZoomon,makesurethatyourmicrophoneismutedduringthebreak.
ADVANCED PROGRAM | ICOICT 2021
15
INSTRUCTIONFORBoFSESSIONHOSTS
• The main responsibilities of a BoF host are opening a live session on Zoom, welcoming
participantstoaBoFsession,andstartingandkeepingtheconversationgoing.
• GeneralguidanceforaBoFsessionhost:
1. Introducethesession,explainthestructureofthesessionandruleofthediscussion.
2. BoF session may include a short presentation from the host to introduce important
questionsorthemesofBoFtopic.
3. SharesomeinterestingquestionsandkeyresearchpapersrelatedtotheBoFtopic.
4. ThemainpurposeofBoFsessionisdiscussionandnetworking.Encourageparticipantsto
sharetheircontactinfoandresearchinterests.
5. BoFsessionscanbeformalorinformalasthehostprefers.
6. Concludethesession.
7. Makesurethetimeisnotviolated
• Organizerwill sendyouan invitation linkviaemail. Justclickon the link to join theZoom
session.
• Pleasejointhesessionatleast10minutesinadvance,pleasetestyourmicrophoneoncejoined
sothatthesessioncanstartontime.
• We recommend that you turn on your video to engage the attendees during the sessional
introduction.
• Asorganizers,wewouldliketoensureasmoothandproductivevirtualconference.
ADVANCED PROGRAM | ICOICT 2021
16
INSTRUCTIONFORATTENDEES
• Youarewelcometoregisterforanysession/eventthatinterestsyouusingregistrationlinks
providedintheicoict.org.Pleasenotethateachroomhaslimitedcapacity.
• Webinarsaredesignedtobein"listen-only"mode,sobydefaultallwebinarattendeesare
mutedbytheorganizer.
• Asorganizers,wewouldliketoensureasmoothandproductivevirtualconference.
• FollowingthevideopresentationthereisashortQ&Asessionmoderatedbythesession
chair.
• Duringthereplayofpresentationvideo,youarewelcometoaskquestionsusingtheQ&A
panel.
• ThemoderatororSessionChairwillselectandaskthequestionsduringtheQ&Asession.We
understandthattheremaybenottimetoaskallthequestions.TheChairwillmakesureto
askcomplementaryquestionsandhopefullyaccordingtothetimetheywerefirstsubmitted.
• Theorganizerwillhavetheabilitytounmuteparticipantsifthisisneededtoelaborateon
theirquestions(pleasenotethatduetothelimitedQ&Aduration,attendeesmaywantto
discusswiththeauthor(s)off-line).
• YoudonotneedtoannounceyourselfarrivingorleavingaSession.
Breaks
Conferencing,onlineandin-personcanbeexhausting!Weneedtotakebreaks.Wewilltake
breaks.Breaksarebuilt-intotheschedule!
• Standupandstretch,getasnack,comebackrefreshed!
• IfyouleaveyourZoomon,makesurethatyourmicrophoneismutedduringthebreak.
ADVANCED PROGRAM | ICOICT 2021
17
KEYNOTESPEAKERS
Dr.Ir.HammamRizaM.Sc,IPU
HeadofAgencyfortheAssessmentandApplicationofTechnology(BPPT)
Profile
Dr. Ir. Hammam Riza currently hold position as DeputyChairman for Natural Resource Technology Development(TPSA) at BPPT and Vice President of ASEAN ChiefInformation Officer Association (*ACIOA). Currentlyundertaking Bigdata and IoT project for Earth and OceanScience Technology including Agricultural productivitymapping (KSA) and Waste to Energy pilot project (PLTSa)
He founded and hold the position of Executive ChiefInformation Officer (CIO) in conjunction with serving asDirector of Center for Information and CommunicationTechnology (PTIK) at the Agency for the Assessment andApplicationofTechnology–BPPT,GovernmentofIndonesia,
deliveringcuttingedge ICT innovativesolution toministries, regionalgovernmentsandstate-ownedenterprises.Title:DigitalInnovationsforPost-PandemicRecoveryAbstractTheCovid-19pandemicisahealthcrisisthataffectsallaspectsofpeople'slives,especiallytheeconomicaspect.WehavenowgottenusedtotheNewNormalera,ofcourse,withahighlevelofvigilance.Allcountriestodayarefacedwiththedilemmaofhowtorestoresocio-economiclife,inthemidstofeffortstostopthespreadofCovid-19.HandlingCOVID-19inIndonesiainvolves3importantrolesfromstakeholders,namely:Government,Society,andTechnologyasenablerstofacilitatetheprocessofTesting,Tracing,Isolating,toTreatment.Thegovernmentcontinuestostrive to prepare some strategies for handling Covid-19, one of which is by establishinginstitutionalsynergiesinthe'TaskForceforResearchandTechnologicalInnovationforHandlingCovid-19 (TFRIC-19) led by the Agency for the Assessment and Application of Technology(BPPT).DigitalTransformationisthekeytonationaleconomicrecoveryandhelpsdeliverthe“newnormal”adaptation.ThisisinlinewiththePresident'smandatetoexpandaccess,improveinfrastructure,acceleratethedigitaltransformationroadmapanditsregulations,includingthepreparation of digital talents. Another aspect that can also encourage increased economicrecovery is the aspect of leadership through synergy and collaboration of all related entities.BPPTandTFRIC-19haveestablishedasystematicandconstructiveecosystemrelatedtotestingtechnology innovations andmedical devices, including: the testing strategy is carried out bydesigningtheRT-PCRtestkitwhichisthegoldenstandardforCovid-19testing,thentheCovid-19Monitorapplicationisdesignedwhichcanmapthemovementofthevirusthroughtracingsuspectsascarriers.Theincreaseintestingneedsisanticipatedbybuildingalevel-2biosafety
ADVANCED PROGRAM | ICOICT 2021
18
mobile labequippedwithcomplete test instruments.Strengthening thepreventionprocess isalsocarriedoutbypreparingappropriatetechnologyfacilitiesandinfrastructureintheformofinnovativehandwashingtoolsanddisinfectantvariants.IntherealmofTreatment,anAI-basedMedical ImageManagementdiagnostic toolhasbeenprepared toprocessX-RayandCT-Scandata,whichcanbeasolutionfortheunevendistributionofhealthtoolsandtechnologythatcanbecome an obstacle in the speed of establishing patient diagnoses under supervision. In thecurativeaspect,especiallyforCovidpatientswithseveresymptomswhorequirerespiratoryaids,3variantsofemergencyventilatorsareprepared,asneeded.Inthefuture,theplanformappingtheentirevirusvarianttoidentifyvirusprofilingwillbecarriedoutthroughpatternrecognition.Thisnationalgenomeresearchwilllaterbecomeavirtualassetforthedevelopmentofbigdataandsupercomputingtobeabletocarryouttheprocessofmonitoringdatacollection,mitigationand produce recommendations and strategic actions that will encourage the emergence ofproduct innovations for handling Covid-19 and awaken national independence in the healthsector. TRFRIC-19 Next generation will continue to be committed through 5 main actions,including:Actiontostrengtheneconomicandtechnologicalstudieswhichincludesupplychainstudies, supply demand studies, pre-commercialization studies, manufacturing industryreadiness studies, TKDN studies, technology audits and innovation ecosystem map studies,Medicaldevicetechnologyinnovationactions,includingICUventilatorinnovations,DirectDigitalRadiography(DDR),quantitativeantibodylevelsmeasuringkitsandAntigenRapidTests,Healthsupplement technology innovation actions, including fermented garlic-based healthsupplements, beta glucan-basedhealth supplements and supplements in the formofnutrientdense biscuits. Actions to strengthen scientific data and applications of artificial intelligence,including AI Application Innovations for Covid-19 detection, Bioprospection Database ofMedicinal Plants, Microbes and Compounds with Potential Drugs for Covid-19 and otherinfectious diseases, as Artificial Intelligence Data Sets, Actions to strengthen Cooperation,CommercializationandMedia.Thisactivityisexpectedtoprovideapositiveestuaryforproductinnovationactionsdevelopedbytheinnovationecosystem.TocreateanindependentIndonesiaandhelpeconomicrecoveryduringtheCovid-19pandemic.
ADVANCED PROGRAM | ICOICT 2021
19
KEYNOTESPEAKERS
Prof.Dr.KazemRahimiProfessorofCardiovascularMedicineatTheUniversityofOxford
Profile
KazemRahimiisaProfessorofCardiovascularMedicineandPopulationHealth,attheUniversityofOxfordandaconsultantcardiologistattheOxfordUniversityHospitalsNHSTrust.Hisresearch interests include hypertension, heart failure,multimorbidityandcardiovascularriskmanagement,usingavariety of methodologies such as individual-patient meta-analysis, large-scale decentralised clinical trials, and digitalhealth technologies. Kazem leads the Deep Medicineprogramme at the Nuffield Department of Women’s andReproductiveHealthwithamajorinterestintheapplicationofmachinelearningapproachestoelectronichealthrecords.Healso leads theBloodPressureLoweringTreatmentTrialists’
Collaboration(BPLTTC),whichisaninternationalcollaborationofallthemajortrialsofbloodpressureloweringdrugs.HeistheDirectoroftheMartinSchoolProgrammeonInformalCitiesandaCo-InvestigatorofthePEAK-Urbanprogramme.Title:MachinelearninganddigitaltechnologiesinthecontextofepidemiologyandclinicaltrialsAbstract Applications of machine learning/AI have given solutions to themost intractable healthcarechallengesanditisalsochangingfundamentallythewaywepracticemedicine.Inthistalk,wediscuss themachine learning from the perspective of large-scale epidemiology. Dataset fromelectronichealthrecordsallowustodiscoverhiddenassociationsbetweendiseases.Inacaseofvalvularheartdisease,itisconsideredasa“degenerative”condition.Bloodpressureisfoundtohaveastrongrelationshipwithvalvularconditions,similartowhathasbeendescribedformajorvascular diseases. The promise of machine learning and digital health is to use analyticalapproachesforhighdimensionaldataset.Machinelearningandhealthcareprofessional’sjobcanbecategorizedintothreetypes,i.e.,Diagnosis,wherewepredictwhatproblemthatthepatientcurrentlyhas,Prognosis,wherewepredictthecourseofdiseaseforapatient,andIntervention,wherewepredicthowapatientwillrespondtoanintervention.Thosearesimilarwithaimsinthemachinelearning;patternrecognition,forecasting,andcausalprediction,respectively.Oneofthe great promises of applyingmachine learning to clinical data is thepossibility of learningoptimal per-patient treatment rules or termed as personalized medicine. There are threecommonmisconceptionsabout‘treatmentfailure’;1.Confusionoflowincidenceconditionswithloweffectivenessof interventions,2.Confusionofdeterministiccausal linkswithprobabilisticmulticause nature ofmost conditions, and 3.Misunderstanding of individual risk and unduecriticism towards average treatment effects. Especially for the third misconception, averagetreatmenteffectsaretheproblemandweneedpersonalizedriskmodelsforbettermanagement.
ADVANCED PROGRAM | ICOICT 2021
20
Individualwilleitherexperienceornotexperiencetheoutcome;so,ariskcannotbedeterminedfortheindividual,modelswillprovideanestimateoftruerisk.Unfortunately,noamountofdataormodelling technique can fix this problem and provide the true risk of an outcome for anindividual,buttheycanbebetteratdiscriminatinggroupsofindividualsbyidentificationofmoremeaningful subgroups or stratification. In this case, we need risk prediction for stratifiedmedicine. When interactions are found they are not reliable and usually only generate ahypothesisforfurthertesting.Thereareopportunitiesformachinelearningforbetterstratifiedmedicine. For low-incident events, risk distribution is skewed, which means that averagetreatmenteffectsmaynotreflecttheexperienceofthetypicalpatient.Moreover,averageeffectsmightbedeterminedbyaminorityofhigh-riskpatients.
ADVANCED PROGRAM | ICOICT 2021
21
KEYNOTESPEAKERS
Dr.DavidTaniarAssociateProfessorDepartmentofSoftwareSystems&Cybersecurity,Monash
University
Profile
DavidTaniarreceivedhisMScandPhDinComputerScience,from Swinburne University of Technology and VictoriaUniversity,respectively.HisresearchisintheareaofBigDataManagement,coveringthe3VsofBigData(Volume,Variety,and Velocity). In Big Data Volume, he works on paralleldatabases,inwhichhehaspublishedabookinthistopic(HighPerformanceParallelDatabaseProcessing,Wiley2008).InBigData Variety, he works on various data structures for datawarehousing, especially for non-relational data. And in BigDataVelocity,heworksonIoTdataprocessing,wherehehascompleted IoT projects for manufacturing, railway,environment and ecology, utility, and healthcare. He haspublishedmore than150 journalpapers invariousareasof
datamanagement.HeistheFoundingEditor-in-ChiefoftwoSCI-Ejournals(DataWarehousingand Mining, and Web and Grid Services). He is currently an Associate Professor at MonashUniversity,Australia.Title:ContactTracingduringCovid-19Pandemic:AnAustralianExperienceAbstractContactTracingistheactivityofretrievinghistoricalactivitiesandtripsforapersonwherehispresenceataspecificlocationmightaffectotherpersonswithinacertainradius.Inrelatedtoacontagiousdisease,aninfectedpersonmightspreadthepathogenstothenearbypeopleduringclosecontactthatcantriggerachainreactionofcommunitytransmission.Thebiggestprobleminobtainingthehistoricalactivitiesinacontacttracingprocedureisprivacyandsecurityissues.Theprivacyissuereferstoprivate-relatedsensitiveinformationthatisnotmeanttobesharedwithanyone.However,duringacontacttracinginvestigation,theauthoritieshavetherighttoknoweverydetailfromasuspectedpatient.Thesecurityissuereferstothesafetyofthesharedprivateinformationtotheauthority.Duetotheseissues,manypatientsarereluctanttosharetheir past activities to the authority. This conditionmakes it even harder to obtain the rightinformationfromthepatients.Thenextconsequenceisthatthespreadingofthediseaseswillbeoff the radar since contact tracing could not be done correctly. Several methods have beenproposedtohelpcontacttracingprocedures.Ingeneral,therearetwotypesofcontacttracingmethods, proximity-based and trajectory-based. While the proximity-based method lackshistoricaltripsandsuffersfrommulti-platformscommunicationissues,trajectory-basedsuffersfromprivacyissues.Thisspeechwilldiscussthesemethodstogetherwiththeirprosandcons.Inconclusion, amethod that canpreserveprivacy and retain the details of the tripwill also beexplainedinthissessionasanalternativemethodtosupportcontacttracing.
ADVANCED PROGRAM | ICOICT 2021
22
KEYNOTESPEAKERS
Prof.HaryadiS.Gunawi,Ph.D
UniversityofChicago
Profile
HaryadiS.GunawiisanAssociateProfessorintheDepartmentof Computer Science at the University of Chicago where heleadstheUCAREresearchgroup(UChicagosystemsresearchon Availability, Reliability, and Efficiency). He received hisPh.D. inComputerScience fromtheUniversityofWisconsin,Madison in 2009. He was a postdoctoral fellow at theUniversity of California, Berkeley from 2010 to 2012. Hiscurrent research focuses on cloud computing reliability andnew storage technology.He has won numerous awardsincluding NSF CAREER award, NSF Computing InnovationFellowship, Google FacultyResearchAward,NetAppFacultyFellowships, and Honorable Mention for the 2009 ACM
DoctoralDissertationAward.His research focus is in improvingdependabilityof storageandcloudcomputingsystemsinthecontextof(1)performancestability,whereinheisinterestedinbuildingstorageanddistributedsystemsthatarerobusttolatencytailsand"limping"hardware,and (2) reliability and scalability, wherein he is interested in combating concurrency andscalabilitybugsincloud-scaledistributedsystems,and(3)interactionsofmachinelearningandsystems, specifically howmachine learning techniques can address operating/storage systemproblems.Title:WhyDoestheCloudStopComputing?Lessons fromThousandsofBugReports, ServiceOutages,andAnecdotalEvidenceAbstract:Cloudcomputing,“thepracticeofusinganetworkofremoteservershostedontheInternettostore,manage,andprocessdata,ratherthanapersonalcomputer,”hasfundamentallychangedtheway societyperformsdailybusinesses and social activities. Emails, text andvideo chats,pictureandvideosharing,blogsandnews,areallbackedbyalargecomplexcollectionofInternetservices,whichwereferas“theCloud.”Asdependencyoncloudcomputing increases,societydemandshighavailability,anideal24/7serviceuptimeifpossible.Yet,serviceoutagesarehardtoescapefrom. Notonlydooutageshurtcustomers, theyalsocause financialandreputationdamages.Minutesofservicedowntimescancreatehundredsofthousandsofdollar,ifnotmulti-million,of loss inrevenue.Company’sstockcanplummetafteranoutage.Sometimes,refundsmustbegiventocustomersasaformofapology.Asrivalsalwaysseektocapitalizeanoutage,millionsofuserscanswitchtoanothercompetitor,acompany’sworstnightmare.Inthistalk,Iwillsharemanystudiesthatmyresearchgrouphasperformedinthelast8years.WeconductedacloudoutagestudyoftensofpopularInternetservicesbyanalyzingover1000headlinenewsandpublicpost-mortemreportsthatdetailhundredsofunplannedoutages.Wealsoconductedacomprehensivestudyofover3000bugreportsinmanypopularopen-sourcecloudsystems.Finally,wealsolookedintofail-slowhardware,anunder-studiedfailuremode,bycollectingover
ADVANCED PROGRAM | ICOICT 2021
23
100anecdotalevidenceoffail-slowhardwareincidentsin12largeinstitutions.Wehopethesestudies canhelp cloud architects and engineers buildmore andmore reliable systems in thefuture.
ADVANCED PROGRAM | ICOICT 2021
24
KEYNOTESPEAKERS
Prof.HadiSusanto
ProfessorattheDepartmentofMathematicsofKhalifaUniversity
Profile
HadiSusantowasbornandgrewupinLumajang(watchthebeautyoftheregionhere),EastJavaprovinceofIndonesia.Hewasanundergraduatestudent(BSc2001)intheDepartmentofMathematics of InstitutTeknologiBandung,with a thesisstudysupervisedbyBarberavandeFliertandEdySoewono.Hesubsequentlydidhispostgraduatestudies(MSc2003,PhD2006) in the Department of Applied Mathematics of theUniversity of Twente under the supervision of Stephan vanGils.HewasVisitingAssistantProfessor (2005-2007)at theDepartmentofMathematicsandStatisticsoftheUniversityofMassachusetts, Amherst mentored by Panos Kevrekidis. HewasthenLecturerinAppliedMathematics(2008-2013)atthe
School of Mathematical Sciences of the University of Nottingham. In 2013, he moved to theDepartment ofMathematical Sciences of theUniversity of Essex as Senior Lecturer and thenProfessorofAppliedMathematics.Heiscurrentlyon-leavefromEssexandisProfessorattheDepartmentofMathematicsofKhalifaUniversity.HeisalsoAdjunctProfessor(GuruBesarLuarBiasa)intheDepartmentofMathematics,InstitutTeknologiBandung.Hisresearchinterestsareintheoreticalandcomputationaldynamicalsystemsandanalysisappliedtothestudyof,amongothers,nonlinearwavesindifferentialequationsthatmodelphysicalreality inmanydifferentfields,includingbiology,condensedmatterphysicsandnonlinearoptics.Healsoenjoyswritingliterature,includingpoetry,proseandessay.Title:COVID-19modellinginIndonesia:Amathematician’sapologyAbstractTheemergenceofSARS-CoV-2inDecember2019inChinaanditsworldwidedisseminationhasbecome amajor public health priority, including in Indonesia. In this talk, I will share somebackgroundstoriesofmyinvolvementintheSIMCOVIDconsortiumthatmodelthediseaseandhelplocalgovernmentsinthecountryinfightingthepandemic.Iwilldiscusssomemathematicalmodels and methods that we developed, that have been used to extract information fordevelopingandevaluatingpolicyresponses.Iwillalsodiscussmy‘apologies’onthelimitationsofthework.
ADVANCED PROGRAM | ICOICT 2021
25
KEYNOTESPEAKERS
RezaKhorshidi,D.Phil.
ProgrammeLead,MachineLearningandBiomedicalInformaticsatTheUniversityof
Oxford
Profile
Dr.RezaKhorshidiiscurrentlythechiefscientistatAIG,andaprincipal investigator(inmachine learningandmedicine)atDeep Medicine program of The University of Oxford. HiscurrentresearchatOxfordisfocusedonprobabilisticmachinelearning, and deep sequence models, for biomedicalinformatics,populationhealth,andprecisionmedicine;morespecifically,heisinterestedinusingmachinelearningforthedevelopment of personalised health predictions andrecommendations, and an improved understanding ofmultimorbidity.Reza’steamatAIG(i.e., InvestmentsAI) isagroup of scientists, engineers, designers and productmanagers/strategists,primarily focusedonthedevelopmentofAI-firstproductsintheFinTechspace.
Title:FromModel.fit() toMarket.fit():Apath towards turningML research intoMLproductswithreal-worldfitAbstractScientistsinthefieldofmachinelearning(ML)–includingdeeplearning(DL)--aspiretobuildbettermodels(usually judgedbybeatingSOTAinwell-definedtasksanddatasets);successfulapplications of such models, on the other hand, are about product-market fit (PMF) inenvironmentswithever-growingcomplexities.AsmanyexpectMLtoplayabiggerroleinoursociety,MLscientists’abilitytoinfluencethisjourneywilldependonputtingMLresearchinaPMFcontextandviceversa(i.e.,optimizingformarket.fit()+⍺*model.fit(),insteadofoptimizingformodel.fit()alone).Therefore,inthistalk,IaimtocoverthegeneralprincipalsofbuildingAIproductsinthe“realworld”,coveringtopicssuchasAIproduct-marketfitandimpactevaluationinmedicine
ADVANCED PROGRAM | ICOICT 2021
26
KEYNOTESPEAKERS
SolehAyubi,Ph.D
ChiefDigitalHealthcareOfficeratBioFarmaIndonesia
Profile
SolehAyubicurrentlytrustedasDirectorofTransformation&DigitalofPTBioFarma(Persero)aswellasChairmanofBUMNMuda.PriortohiscareerinSTATE-OWNEDenterprises,Ayubiwas a business, tech, & healthcare professional, who had avariety of prestigious experiences in various internationalhealthcompaniesintheUnitedStates.SolehgraduatedfromtheInformaticsEngineeringProgramin2005,thengraduatedfromtheInformaticsMasterProgramin2007,bothfromtheBandung Institute of Technology. Soleh continued hiseducationandobtainedaDoctorofPhilosophy(Ph.D)degreeinHealthSciencesfromtheUniversityofPittsburghin2012.Before returning to Indonesia and joining Bio Farma, Soleh
had 12 years of global experience in healthcare by joining several international companies;amongothers,NovoNordisk(Seattle,UnitedStates)asDirector,HeadofDigitalTherapeuticsandData Science for 2019-2020; Unitedhealth Group (Boston, United States) as Director ofInnovation, Research and Development in 2017 - 2019; Boston Children's Hospital (Boston,United States) as Technical Innovation Manager 2013-2016; Veterans Affairs Hospital(Pittsburgh, USA) as Senior Software Engineer in 2011-2012. Besides in theworld of healthindustry,SolehAlAyubiisalsoactiveintheworldofeducationandwasappointedtobecomeDigitalandHealthEntrepreneurshipatHarvardUniversity(Boston,UnitedStates)since2017untilnow,whichhasthetaskofprovidingtrainingandmentoringtomedicalstudents,doctoral(PhD)andpost-doctoralstudentsatHarvardMedicalSchool.Since2020,hehasalsobecometheAdjunctFacultyattheSchoolofBusinessManagement,BandungInstituteofTechnology.Inthescientific andR&Dcommunities, besidesholding2patents inAmerica, Soleh is also active invarious international conferences and publications in the fields of healthcare innovation andhealthcaretechnologybothasaspeaker,reviewer,andasajournalwriter.Asawriter,Solehhaspublished15papersininternationaljournalsandproceedings.Title:Covid-19VaccineDistribution:anexampleofAcceleratinghealthcaretransformationAbstractIndonesia is severely impactedbyCOVID-19withmore than twomillion cases anda currentthousanddeathsaday.Witha270millionpopulationlivingin514citieswithin6,000islands,itisalogisticalnightmareforvaccinedistribution.TonavigatethatchallengeBioFarmaasastate-ownedenterprisespecializedinvaccineproductionanddistributionhasbeencollaboratingwithseveraltechnologycompaniescoordinatedbyTelkomtodevelopaSupplyChain4.0systemofCOVID-19vaccinedistribution.
ADVANCED PROGRAM | ICOICT 2021
27
Ablessingindisguise,the3-yearstalleddevelopmentoftheSupplyChain4.0calledTrackandTracesystem,wascatalyzedbythispandemic.Thesystemenablesustotrackandtraceeveryvialofvaccinefromthefirstmileofproductionpipelinetothelastmiledeliveryattheclinicorhospital.Thesystemappliesa2-DDataMatrixinvaccinevial,secondarypackage,andtertiarypackagewhich allows overall product identification, traceability, and authentication from end-to-end.UtilizingIoTsensors,thesystemsteadilyrecordsthelocationandtemperatureofvaccineswhilebeingstoredortransportedtoensurevaccinequalityuntilthelastmile.Thesystemisintegrablewithothermonitoringsuchasseniorleader’sdashboardattheMinistryofHealth,theMinistryof State-ownedEnterprises, andother senior leaderoffices.This allows forwell-orchestratedoperations,fastresponsetothechallengeinoperationsanddata-drivendecisionmakingforbothoperationsandstrategies.
ADVANCED PROGRAM | ICOICT 2021
28
KEYNOTESPEAKERS
Dr.MohamadHardymanBarawi
UniversitiMalaysiaSarawak
ProfileHardyman is currently an academician at the UNIMAS. HereceivedhisPhDinComputingSciencefromtheUniversityofAberdeen. Before joining UNIMAS, he was an OperationSupportSpecialistatHewlettPackardandaSystemEngineeratMiszaTechnology.He is best known for his work on sentiment analysis andopinion mining, automatically labelling sentiment-bearingtopics, abstractive and extractive labelling, and opinionsummarization.Overthelastfewyears,hehasbeeninterestedinstatistical techniques fornatural languagegeneration.Hisresearchinthisareahasincludedworkinthesubareasoftext
summarization, topic modelling, and, more recently, opinion summarization through wordstatistics,linearapproach,andlexicalresourceacquisitionthroughstatisticalmeans.Title:Modelling Sentiment and Contrastive Opinion of COVID-19 Pandemic on Social Media:LinkingComputerScienceandSocialScienceAbstractSentiment analysis is the computational study of people’s opinions, attitudes, and emotionsexpressedinatextorwrittenlanguage.Duetomanychallengingresearchproblemsandawiderangeofpracticalapplications, ithasbecomeoneof themostactiveresearchareas innaturallanguageprocessing inrecentyears. In this talk, Iwilldiscussmainstreamsentimentanalysisresearch before moving on to describe some more recent work on modelling opinion andcommentsofCOVID-19PandemiconSocialMedia.Thisresearchnaturallyconnectscomputerscience and social science, especially communication and political sciences, in social mediaanalysis.
ADVANCED PROGRAM | ICOICT 2021
29
KNOWLEDGETRANSFER
WuShiwei
HUAWEICLOUD,CTOofAPACRegion
Profile
15 years working in the IT industry, I have experiencedvariousroles fromdeveloperto leadsolutionarchitect,withvariousprojectsinvariousenvironmentsfromastartupofadozenpeopletoatop100MNC.AsCTOofHuaweiCloud,APACregion,oneofmykeyresponsibilities is to lead thesolutionteam to address technical challenges from variousprojects/clients with proper solution designs; advise R&Dteamregarding toproduct/serviceenhancementsaswell assupportbuildingpartnereco-systemofHuaweiCloudinthisregion.
Title:Workforce Enablement in the New Normal of Post-Pandemic, Powered by CloudTechnologyAbstractTheoutbreakofpandemicwasquiteunexpected.Manydigitalsolutions/innovationswereputinto place for emergency use only, without considering the long term impact, especially insupportingworkforce;however it isbecomingincreasingly likelythatwewillbe livinginthisnewnormalthusweneedtorethink,redesigntheinitialsolutionstoenhancesecurity,enrichfeaturesandimproveuserexperience.Inthissession,Iwillexploresometypicaldigitalsolutionsforworkforceenablementanddiscusshowthesetypesoftechnologymayimpactourworkandlifeandwhatfeaturesthatneedtobeenabledtobecomemoresustainable.
ADVANCED PROGRAM | ICOICT 2021
30
TUTORIALSPEAKERS
Dr.BayuErfianto
SeniorLecturerandresearcheratSchoolofComputingTelkomUniversity
ProfileBayuErfiantograduatedfromtheDepartmentofMathematics,majoring in Computer Sciences, Universitas Padjadjaran(UNPAD), Bandung, Indonesia in 1999 for his Bachelor ofSciences(S.Si.).HecontinuedhisMasterDegreeinTelematics,UniversiteitTwente,theNetherlandsfrom2002to2004.HealsocontinuedhisMasterDegreeinInformationTechnology,UniversitiTeknologiPetronasMalaysiafrom2007to2009.Hereceived Doctoral Degree in Electrical Engineering fromSchool of Electrical Engineering and Informatics (STEI),Institut Teknologi Bandung in 2017. Dr. Bayu was also asPostdoctoral researcher in School of Electrical EngineeringandInformatics(STEI)in2017-2018.Bayuiswiththeschool
ofcomputing(FakultasInformatika),TelkomUniversity,Bandung,IndonesiaasLecturersince2002 (a.k.a. STT Telkom). His research interests are in the field of Cyber-physical System,including:InternetofThings,NetworkandControlSystemofCyber-physicalSystem,andFormalModellingofCyber-physicalSystem.Title:Chest-wallMotionTrackingusingInertialMeasurementUnits(IMUs)AbstractInertialMeasurementUnits(IMUs) ismostlyembedded inwearabledevices.Nowadays, IMUsmotion tracking systems are allowing for long-lasting tracking of user motion in a situatedenvironment.Insteadofbodymotiontracking,ontheotherhand,wearingIMUsonthechestwalloffers a few advantages, such as for cardiac activity parameters estimation and respirationparameterestimation.Forexample,currentresearchshowsthatinertialsensorsarelow-costandeasy-to-use breathing-monitoring systems. Breathing parameters from chest-wall inclinationsignalsareeasilymeasuredusingIMU.ThistutorialpresentsseveraltechniquesforIMUsbasedmotion tracking to reconstruct chestwallmotionwith respect to angular velocity and linearacceleration.Severaltechniquesfordataorsignalprocessingarchitecturesarealsodiscussed.Thistutorialalsotriestointroducetheapplicationsofchest-wornIMUsbasedmotiontrackingtoestimateheart-rate,blood-pressure,andrespirationrate.
ADVANCED PROGRAM | ICOICT 2021
31
TUTORIALSPEAKERS
DhanySaputra,Ph.D
DataScientistatBrüel&KjærVibro
Profile
Dr.DhanySaputraisabioinformaticiananddatascientist.HecurrentlyworksatBrüelandKjærVibro.HefinishedhisPh.D.inBioinformaticsattheCenterforGenomicEpidemiology,TUDenmark,in2013.Hehasparticipatedinmanybioinformaticscases, from controlling outbreaks, paleontology excavationanalysis, stool transplant, marine biology, food safety, andcancer research. He has anM.Sc. in Computer Science fromUniversitiTeknologiPETRONASMalaysiain2008andaB.Sc.in Information System from Sepuluh Nopember Institute ofTechnologySurabayaIndonesiain2005.
Title:BioinformaticshasHelpedPrevent&ControlaPandemic.What'sNext?AbstractBioinformaticsandcloudtechnologieshavebeenbeneficialinmanycountries'decision-makingduring this pandemic. Sequence analysis algorithms, such as fast algorithms on sequencealignment,de-novoassembly,geneannotation,geneexpressionanalysis,geneprediction,andantibiotic resistance finding, have been crucial in solving an outbreak. Nowadays, DNAsequencinghasbecomearoutinetask,andpeoplecanevendoitintheirkitchen.ImprovementsinDNAsequencingtechniqueshaveencouragedthepopularityofshotgunmetagenomics.Thewhole metagenome shotgun enables finding the microbial diversity - or in medicalbioinformatics,thelistofpathogens-inonesample.Additionally,itisalsotheprimarykeytodiscover the cure for genetic diseases and cancer. Together with microbiome research,bioinformaticshasencouragedfecalmicrobiotatransplant(FMT)researchtocuremanydiseasessuchasCDI,bipolardisorder,autism,andcancer.
ADVANCED PROGRAM | ICOICT 2021
32
TUTORIALSPEAKERS
Assoc.Prof.Dr.Md.ShohelSayeed
AssociateProfessorandresearcheratMultimediaUniversity(MMU)
Profile
Dr. Shohel has more than 25 years meritorious workingexperienceandheholdsachallengingcareerwhichcombinesresearch,versatileadministrationandexcellentteaching.Hiscore research interest is in the area ofBiometrics, big data,cloudcomputing, artificial intelligence, information security,image and signal processing, pattern recognition andclassification. He has published over 60 research papers ininternational peer-reviewed journals and internationalconferenceproceedingsasaresultofhis researchwork.Dr.Shohel has been a member of Multimedia University since2001 and now he serves as an Associate Professor and theChairpersonofCentreforIntelligentCloudComputing(CICC)
oftheFacultyofInformationScienceandTechnology.HereceivedhisdoctorofphilosophyfromMultimedia University in Engineering, specializing in hand signature verification, and holdsmastersofinformationtechnologydegreefromtheUniversityKebangsaanMalaysia,specializinginindustrialcomputing.HereceivedhisBachelorofSciencedegreeinAgriculturalsciencefromBangladeshAgriculturalUniversity.Title:BigData:Trends,Challenges&OpportunitiesAbstractEveryday,approximately2.5quintillionbytesofdataarecreated.Thesedatacomefromdigitalpictures,videos,poststosocialmediasites,intelligentsensors,purchasetransactionrecords,cellphoneGPSsignals,tonameafew.ThisisknownasBigData.ThereisnodoubtthatBigDataandespeciallywhatwedowithithasthepotentialtobecomeadrivingforceforinnovationandvaluecreation.Innovationsintechnologyandgreateraffordabilityofdigitaldeviceshavepresidedovertoday’sAgeofBigData,anumbrellatermfortheexplosioninthequantityanddiversityofhighfrequencydigitaldata.Thesedataholdthepotentialasyet largelyuntappedtoallowdecisionmakerstotrackdevelopmentprogress,improvesocialprotection,andunderstandwhereexistingpolicies and programmes require adjustment. Turning Big Data—call logs, mobile-bankingtransactions, online user-generated content such as blog posts and Tweets, online searches,satellite images, etc. into actionable information requires using computational techniques tounveil trendsandpatternswithinandbetweentheseextremely largesocioeconomicdatasets.Newinsightsgleanedfromsuchdataminingshouldcomplementofficialstatistics,surveydata,and information generated by EarlyWarning Systems, adding depth and nuances on humanbehavioursandexperiencesanddoingsoinrealtime,therebynarrowingbothinformationandtimegaps.Thepromiseofdata-drivendecision-making isnowbeing recognizedbroadly, andthereisgrowingenthusiasmforthenotionof``BigData.’’Thereiscurrentlyawidegapbetweenitspotentialanditsrealization.
ADVANCED PROGRAM | ICOICT 2021
33
Heterogeneity, scale, timeliness, complexity, and privacy problems with Big Data impedeprogressatallphasesofthepipelinethatcancreatevaluefromdata.Alargeamountofdatatodayisnotnativelyinstructuredformat;forexample,tweetsandblogsareweaklystructuredpiecesof text, while images and video are structured for storage and display, but not for semanticcontentandsearch.Transformingsuchcontent intoastructured format for lateranalysis isamajor challenge.Thevalueofdata explodeswhen it canbe linkedwithotherdata, thusdataintegration isamajorcreatorofvalue.Sincemostdata isdirectlygenerated indigital formattoday,wehavetheopportunityandthechallengebothtoinfluencethecreationtofacilitatelaterlinkageandtoautomaticallylinkpreviouslycreateddata.Dataanalysis,organization,retrieval,andmodelling are other foundational challenges. Data analysis is a clear bottleneck inmanyapplications,bothduetolackofscalabilityoftheunderlyingalgorithmsandduetothecomplexityofthedatathatneedstobeanalyzed.Finally,presentationoftheresultsanditsinterpretationbynon-technical domain experts is crucial to extracting actionable knowledge. Themany novelchallengesandopportunitiesassociatedwithBigDatanecessitate rethinkingmanyaspectsofthesedatamanagementplatforms,whileretainingotherdesirableaspects.ItshouldbepointoutthattheappropriateinvestmentinBigDatawillleadtoanewwaveoffundamentaltechnologicaladvancesthatwillbeembodied inthenextgenerationsofBigDatamanagementandanalysisplatforms,products,andsystems.Thus,weshouldbelievethattheseresearchproblemsarenotonlytimely,butalsohavethepotentialtocreatehugeeconomicvalueintheworldeconomyforyears to come.However, they are also hard, requiring us to rethink data analysis systems infundamentalways.AmajorinvestmentinBigData,properlydirected,canresultnotonlyinmajorscientificadvances,butalso laythefoundationforthenextgenerationofadvances inscience,medicine,andbusiness.
ADVANCED PROGRAM | ICOICT 2021
34
TUTORIALSPEAKERS
MohEdiWibowo,S.Kom.,M.Kom.,Ph.D.
SeniorLecturerandresearcheratGadjahMadaUniversity(UGM)
Profile
Moh EdiWibowo received the bachelor degree andmasterdegreeincomputersciencefromUniversitasGadjahMadain2004and2006respectively,andthePh.D.degreeincomputerscience fromQueenslandUniversity of Technology in 2014.His current focus area is human and object detection andidentification. He currently works as the vice head of thedoctoralprogramofcomputerscienceat theDepartmentofComputerScienceandElectronics,UniversitasGadjahMada.
Title:FaceAnalysisinPublicSpaceAbstractThistutorialdiscussesvariousmethodstodetectandtoidentifyfaces,faciallandmarks,andfacialattributesinimagesandvideoscollectedfrompublicspaces.Suchmethodsprovideinformationthatmight be useful for public spacemonitoring and access control in particular during thecurrent pandemic situation. Algorithms such as cascaded regression and probabilisticdiscriminantanalysiswillbediscussedandimplementedinthetutorial.
ADVANCED PROGRAM | ICOICT 2021
35
TUTORIALSPEAKERS
JuneidiTsai
TechnicalSpecialistatMicrosoft
ProfileCRMSpecialistwithmorethan10yearsofprofessionalworkexperience encompassing various facets of application anddata analysis across many industries including automotive,education,retail,andfinancialserviceindustry.
Title:MicrosoftEducationTransformationFrameworkforHigherEducationAbstractHighereducationinstitutionsarecomplexandmultifacetedorganizationscomprisedofmultipledepartments thatmustwork together toexecutea successfulvision.TheMicrosoftEducationTransformationFrameworkenablesaholisticlookattheinstitutionbutprovidesyouwiththeabilitytodevelopyourdigitalstrategyindiscretephases,answeringthatallimportantquestion,“WhereshouldIstart?”The Microsoft Education Transformation Framework provides practical advice to help youdevelop a strategy for digital transformationwith a holistic, long-term view implemented indiscretephasesthatyoucanbegintodaytohelpdriveinnovativestudentengagement,transformoperations,andensureasecure,connectedcampustoempoweryoutoreimagineeducation.
ADVANCED PROGRAM | ICOICT 2021
36
SOCIALEVENTS
BirdsofafeatherSessions
Host:Afiahayati,Ph.DUniversitasGadjahMada
Theme:AIandBioinformaticsforCOVID-19
Afiahayati,S.Kom.,M.Cs,Ph.DisaLecturerandResearcherattheDepartmentofComputerScienceandElectronics,FacultyofMathematicsandNaturalSciences,GadjahMadaUniversity.ShefinishedherDoctoralDegreeinDept.ofBiosciencesandInformatics/Faculty of Science and Technology, KeioUniversity, Japan. Her research interests includeBioinformatics,MachineLearningandArtificialIntelligence.
Host:Dr.MuhammadJohanAlibasaTheme:DigitalfootprintsduringCOVID-19pandemics
Muhammad Johan Alibasa is a lecturer at the School ofComputing, T elkom University, Indonesia. He received theSchool of Computer Science Research Students ExcellencePrize2019fromtheUniversityofSydney,Australia,wherehecompletedhisdoctoraldegree.HewasalsoarecipientoftheIndonesian Endowment Fund for Education (LPDP)scholarship fromMinistryofFinance,Republicof Indonesia.His research interests include human-computer interactionand artificial intelligence, specifically on the applications ofmachinelearningtoimprovewellbeing.PriortohisPhDstudy,heworkedasamobileapplicationdeveloperanddevelopedmobileappsintheareaofworkplacewellbeing.Hereceived
theM.Sc.andB.Sc.degreesinelectricalengineeringfromInstitutTeknologiBandung,Indonesia,in2014and2012.
ADVANCED PROGRAM | ICOICT 2021
37
SOCIALEVENTSBirdsofafeatherSessions
Host:KemasRahmatShaleh,Ph.D
TelkomUniversityTheme:KnowledgeManagement,KnowledgeRepresentationandreasoning
KemasRahmatSalehWiharjaisafulltimelecturerinTelkomUniversity since 2006. He got his Ph.D from University ofAberdeen in2020andhis research interestsareknowledgerepresentation and reasoning, knowledge graph, and datascience.
Host:Dr.AgusHartoyoTelkomUniversity
Theme:ArtificialIntelligenceofThings(AIoT):TheintersectionofAIandIoT
AgusHartoyoreceivedaB.Sc.inInformaticsfromITTelkom(nowTelkomUniversity),Indonesia,in2008onhisthesisinAIdevelopmentforsolvingaprobleminNLP.Hecompletedhis M.Sc. in Computer Science from TU Kaiserslautern,Germany,in2015withhisresearchworkfocusingonsolvinga reachability problem in the safety and reliability ofembedded systems. In 2021 he obtained a Ph.D. inBioinformatics from Swinburne University of Technology,Australia,ona thesis in the inferenceandmodelingofbraindynamics.HisPh.D.dissertationwasexaminedbytwoexpertsincomputationalbiology,oneofwhomgradedthethesiswiththehighestgrade:“Excellent-SummacumLaude”.Rightafter
hisPh.D.,AgusHartoyo joined theSchoolofComputing,TelkomUniversity, as a lecturer andresearcher.Hisresearchinterestsincludeartificialintelligence,statisticalinference,safetyandreliabilityofembeddedsystems,bioinformatics,andcomputationallinguistics.
ADVANCED PROGRAM | ICOICT 2021
38
SOCIALEVENTSBirdsofafeatherSessions
Host:Dr.MayaAriyantiTelkomUniversity
Theme:ELearning
Dr.MayaAriyanti SE.,MM;Permanent lecturerofMasterofManagement Study Program, Faculty of Economics andBusiness,TelkomUniversitysince2008.CompletedBachelorofManagement inMarketingManagement Concentration attheFacultyofEconomics,ParahyanganUniversity,Bandung,1996, then continued to Master of Management (MM)Concentration inMarketingat theFacultyofEconomicsandBusiness, Padjadjaran University in in 1999 and Doctor ofEconomicswithaconcentrationinMarketingManagementatPadjadjaranUniversity-Bandungin2009.Hisfieldofstudyupto now is Marketing Management, Services and DigitalMarketing.Hasseveralarticlespublishedinreputablejournals
forthefield,haspublishedseveralbookssuchasCreditManagement,IntroductiontoMonetaryTheory, andMarketing Plan. The author is also a consultant andwrites cases in the field ofmarketing management. Currently the author is also a lecturer for management courses,introductiontobusiness,marketingmanagement,digitalmarketing,digitalconsumerbehavior,digitalbusinessconsumerbehavior,anddigitalmarketingcommunications.
ADVANCED PROGRAM | ICOICT 2021
39
TECHNICALPARALLELSESSIONSCHEDULEParallelSession#1:Wednesday,August4,2021,11:00-13:00UTC/GMT+7
Session 1A–ApplicationsforPost-PandemicRecovery
Venue RoomA:Yudhistira
SessionChair Dr.AgusSihabuddin,S.Si.,M.Kom
Time PaperTitle Presenter
11:00TheroleoftechnologycapabilitiesandinnovationcapabilitiesinachievingbusinessresilienceofMSMEsduringCovid-19:EmpiricalStudy
GrisnaAnggadwita
11:20Gaze-ControlledDigitalSignageforPublicHealthEducationduringCovid-19Pandemic SunuWibirama
11:40AnalysisoftheHouseofRisk(HOR)ModelforRiskMitigationoftheSupplyChainManagementProcess(CaseStudy:KPBSPangalenganBandung,Indonesia)
RatihHendayani
12:00EvaluationoftheSocialRestrictionanditsEffecttotheCOVID-19SpreadinIndonesia
InnaSyafarina
12:20ContributingClinicalAttributestoCOVID-19MortalityinJakarta:MachineLearningStudy
MuhamadErzaAminanto
Session 1B–ComputerVision
Venue RoomB:Bima
SessionChair TeeConnie(MMU)
Time PaperTitle Presenter
11:00Multi-TargetRegressionUsingConvolutionalNeuralNetwork-RandomForestforEarlyEarthquakeWarningSystem
AdiWibowo
11:20 Vision-BasedEmployeeActivityClassification RizalPutra
11:40CompressiveSensingImageWatermarkingOrthogonalMatchingPursuit
IrmaSafitri
12:00TrafficSignRecognitionwithConvolutionalNeuralNetwork
ZhongBoNg
12:20DeepConvolutionalGenerativeAdversarialNetworkApplicationinBatikPatternGenerator
Moch.ChamdaniMustaqim
ADVANCED PROGRAM | ICOICT 2021
40
Session 1C–DataScience
Venue RoomC:Arjuna
SessionChair Dr.YuliantSibaroni
Time PaperTitle Presenter
11:00ConvolutionalNeuralNetworksforIndonesianAspect-BasedSentimentAnalysisTourismReview AhmadHidayatullah
11:20SentimentAnalysisonMarketplaceReviewusingHybridLexiconandSVMMethod
MuhammadMukhtar
11:40ForecastingNumberofCOVID-19CasesinIndonesiawithARIMAandARIMAXModels
BimoAji
12:00DisasterTweetClassificationBasedOnGeospatialDataUsingtheBERT-MLPMethod IqbalMaulana
12:20CyberbullyingDetectiononIndonesianTwitterusingDoc2VecandConvolutionalNeuralNetwork
ShindyLaxmi
Session 1D–E-LearningandHCI
Venue RoomD:Nakula
SessionChair Dr.KusumaAyuLaksitowening
Time PaperTitle Presenter
11:00DigitalNudgeEvaluationonCOVID-19tracingApplication
DyahSukmaningsih
11:20DevelopingSuicideRiskIdeaIdentificationforTeenager(SERIINA)MobileAppsPrototypeusingExtendedRapidApplicationDevelopment
TeniaWahyuningrum
11:40DesigningAnEducationalGameEvaluationFrameworkBasedOnGameMechanic
FaisZharfanAzif;SatrioRukmono
12:00CultivatingRecyclingAwarenessinPreschoolersusingAnimatedInteractiveComic
SitiZulaihaAhmad
12:20AnalysisInfluenceofTheOrganizationalLearningEnvironmentFactorsToEncourageEmployeeMotivationUsingE-Learning
NanikQodarsih
ADVANCED PROGRAM | ICOICT 2021
41
Session 1E–Healthcare,Bioinformatics,andBiomedicalApplications
Venue RoomE:Sadewa
SessionChair Dr.DodiQory
Time PaperTitle Presenter
11:00DistributedPhylogeneticTreeProcessingonBiologySequencesUsingMapreduce RenaningSusilo
11:20LinearRegressionModeltoPredicttheSpreadofCOVID-19inTangerangCity
YusufSudiyono
11:40StrategicInformationSystemPlanningforIndonesiaNon-franchisePharmaciesBasedonJohnWardandFactorAnalysisMethod
TabahArwiyanto
12:00FlexibleMulti-LayerConduraFabricUltraWide-BandAntennaForTelemedicineApplication
YusnitaRahayu
12:20AnonymizingPrescriptionDataAgainstIndividualPrivacyBreachinHealthcareDatabase
DediGunawan
ADVANCED PROGRAM | ICOICT 2021
42
TECHNICALPARALLELSESSIONSCHEDULEParallelSession#2:Wednesday,August4,2021,14:00-16:00UTC/GMT+7
Session 2A–Networking,IoT,andSecurity
Venue RoomA:Yudhistira
SessionChair Dr.BayuErfianto
Time PaperTitle Presenter
14:00SimulationOfJellyfishTopologyLinkFailureHandlingUsingFloyd-WarshallandJohnsonAlgorithminSoftwareDefinedNetworkArchitecture
MuhammadNugroho
14:20GameTheoreticalPowerControlinHeterogeneousNetwork AnggunIsnawati
14:40IoTDroneCameraforaPaddyCropHealthDetectorwithRGBComparison
AjiPutrada
15:00 AReviewonIoTwithBigDataAnalytics AbuFuadAhmad
15:20VehicleBlindSpotAreaDetectionUsingBluetoothLowEnergyandMultilateration
AjiPutrada
Session 2B–ComputerVision
Venue RoomB:Bima
SessionChair LimKianMing
Time PaperTitle Presenter
14:00FacialEmotionRecognitionusingTransferLearningofAlexnet
SarmelaRajaSekaran
14:20VisuallySimilarHandwrittenChineseCharacterRecognitionwithConvolutionalNeuralNetwork
WeiHanLiu
14:40PneumoniaClassificationusingGabor-ConvolutionalNeuralNetworksandImageEnhancement
MuhammadAlfarizy
15:00FingerprintEnhancementusingIterativeContextualFilteringforFingerprintMatching
BramaSatria
15:20HistogramofOrientedGradientRandomTemplateProtectionforFaceVerification
LucasChongWeiJie
ADVANCED PROGRAM | ICOICT 2021
43
Session 2C–DataScience
Venue RoomC:Arjuna
SessionChair Dr.ImeldaAtastina
Time PaperTitle Presenter
14:00Aspect-BasedSentimentAnalysisinBeautyProductReviewsUsingTF-IDFandSVMAlgorithm NadiraArthamevia
14:20AspectTermExtractionUsingDeepLearning-BasedApproachonIndonesianRestaurantReviews
RachmansyahAdhiWidhianto
14:40 SpamDetectiononIndonesianBeautyProductReviewMuhammadAhsan
Athallah
15:00EmotionClassificationonIndonesianTwitterUsingConvolutionalNeuralNetwork(CNN) FirhanMaulanaRusli
15:20MappingComplexTouristDestinationPreferences:NetworkPerspectives
DianRamadhani
Session 2D–E-LearningandHCI
Venue RoomD:Nakula
SessionChair DanaSulistyoKusumo,Ph.D
Time PaperTitle Presenter
14:00ExploringtheexistenceandvariationofGamePlayerTraitsamongUndergraduatestudentsinMalaysia
MageswaranSanmugam
14:20ImplementationofContinuousIntegrationandContinuousDelivery(CI/CD)onAutomaticPerformanceTesting
MohammadRizkyPratama
14:40UnderstandingGovernmentReorganizationImpactfromKnowledgeManagementPerspective:AStudyCase
YuliaSulistyaningsih
15:00EnterpriseResourcePlanningTeachinginPostPandemicusingGamification
KaushalJheengut;DineshaCauleechurn;
BibiCadersaib
15:20 RPA-basedBotsforManagingOnlineLearningMaterials SitiFatimahAbdulRazak
ADVANCED PROGRAM | ICOICT 2021
44
Session 2E–Healthcare,Bioinformatics,andBiomedicalApplications
Venue RoomE:Sadewa
SessionChair SatriaMandala,Ph.D
Time PaperTitle Presenter
14:00Holick'sRuleImplementation:CalculationofProducedVitaminDfromSunlightBasedonUVIndex,SkinType,andAreaofSunlightExposureontheBody
JonathanSalomo
14:20T-COFFEEMultipleSequenceAligneronHadoopSparkCluster
ViebiyantyPrihatiningrum
14:40RelaxationOscillatorUsingClosed-loopDualComparatorforBiomedicalApplications
TheodoraValerie
15:00 WirelessProgrammablebodysensornetworksandSituatedHealthcare
AlbertoFaro
ADVANCED PROGRAM | ICOICT 2021
45
TECHNICALPARALLELSESSIONSCHEDULEParallelSession#3:Thursday,August5,2021,08:15-10:15UTC/GMT+7
Session 3A–Networking,IoT,andSecurity
Venue RoomA:Yudhistira
SessionChair Dr.VeraSuryani
Time PaperTitle Presenter
08:15DetectionofSinusoidswithFrequencyDriftinWhiteGaussianNoise
BradleyComar
08:35ToneDetectionSystemDesignforTargetswithFrequencyDrift
BradleyComar
08:55AADC3:Active-ActiveDistributedControllerwith3-in-1AsynchronousHeartbeatSynchronizationMethodinSoftware-DefinedNetworks
MuhammadNugroho
09:154GLTECellularNetworkCoveragePlanningandSimulationonMandalayAreawithPropagationModelCost-Hatta
AhmadIdris
09:35USBFlashDrivesForensicAnalysistoDetectCrownJewelDataBreachinPT.XYZ(CoffeeShopRetail-CaseStudy)
DanielSeptianto
Session 3B–ComputerVision
Venue RoomB:Bima
SessionChair Dr.EmaRachmawati
Time PaperTitle Presenter
08:15EnhancedAlexNetwithSuper-ResolutionforLow-ResolutionFaceRecognition
JinChyuanTan
08:35AnEnd-to-EndOpticalCharacterRecognitionPipelineforIndonesianIdentityCard
AndreasChandra
08:55 AStudyofBatikStyleTransferusingNeuralNetwork AdityaIhsan
09:15TomatoPlantDiseaseIdentificationthroughLeafImageusingConvolutionalNeuralNetwork
AuliaYoren
ADVANCED PROGRAM | ICOICT 2021
46
Session 3C–DataScience
Venue RoomC:Arjuna
SessionChair DiditAditya,Ph.D
Time PaperTitle Presenter
08:15Non-StationaryOrderofVectorAutoregressioninSignificantOceanWaveForecasting FikkaRaudiya
08:35FN-Net:ADeepConvolutionalNeuralNetworkforFakeNewsDetection
KianLongTan
08:55SentimentAnalysisofOjekOnlineUserSatisfactionBasedontheNaïveBayesandNetBrandReputationMethod
AlamRahmatulloh
09:15RawPaperMaterialStockForecastingwithLongShort-TermMemory
FebryoKurniawan
09:35MobileCustomerBehaviourPredictiveAnalysisforTargetingNetflixPotentialCustomer
SuryadiTanuwijaya
Session 3D–E-LearningandHCI
Venue RoomD:Nakula
SessionChair SitiZainabIbrahim
Time PaperTitle Presenter
08:15MasterDataManagementMaturityModel(MD3M)Assessment:ACaseStudyinSecretariatofPresidentialAdvisoryCouncil
ChielsinKo
08:35CapturingInstitutionandLearnersReadinessofe-LearningImplementation:ACaseStudyofaUniversityinBandung,Indonesia
DawamDwiJatmikoSuwawi
08:55SatisfactionFactorsofIndonesianNationalCivilServantRecruitmentSystem
GalihAvianto
09:15ImplementationandAnalysisofReusabilityFrameworkDesignforEventUserInterfaceComponentinPhaser3 DanaSulistyoKusumo
09:35ThePreliminaryStudyonthePerceptionofEngineeringStudentsonBlendedLearning
MinChiLow
ADVANCED PROGRAM | ICOICT 2021
47
Session 3E–Healthcare,Bioinformatics,andBiomedicalApplications
Venue RoomE:Sadewa
SessionChair NgChongHan
Time PaperTitle Presenter
08:15ImplantSegmentationinRadiographicImageryUsingMultiresolutionMTANNandWaveletDecomposition RanggaPerwiratama
08:35ImprovingMulti-ClassMotorImageryEEGSignalsClassificationUsingEnsembleLearningMethod
DeniNugroho
08:55ImplementationandExperimentalCharacterizationofDual-BandWearableReflectorComposedofAMCStructureforWirelessCommunication
IchsanNusobri
ADVANCED PROGRAM | ICOICT 2021
48
TECHNICALPARALLELSESSIONSCHEDULEParallelSession#4:Thursday,August5,2021,10:15–12:15UTC/GMT+7
Session 4A–Networking,IoT,andSecurity
Venue RoomA:Yudhistira
SessionChair Dr.RioGunturUtomo
Time PaperTitle Presenter
10:15ModifiedPixelValueOrdering-basedPredictorforReversibleDataHidingonVideo
TohariAhmad
10:35SimulationandAnalysisofPartialTransmitSequenceonPalmDateLeafClippingforPAPRValueReduction
VincentVincent
10:55 DesignAutomationofSinglePhotonCountingMethodforQuantumRandomNumberGeneration
DedyPutranto
11:15OntheModificationsofaDigitalSignatureAlgorithmwithSecretSharing
MaretaWahyuArdyani
11:35ConnectedVehicleCommunicationConceptforFloodLevelWarningUsingLowCostMicrocontroller
SumendraYogarayan;MohdFikriAzli
Abdullah
11:55Randomness,Uniqueness,andSteadinessEvaluationofPhysicalUnclonableFunctions
RivaldoSembiring
Session 4B–ComputerVision
Venue RoomB:Bima
SessionChair WikkyFawwaz,Ph.D
Time PaperTitle Presenter
10:15ALow-CostHigh-AccuracyThermalCameraUsingOff-the-shelfHardwareDevices
Dinh-TienTran
10:35Sentinel1ClassificationforGarlicLandIdentificationusingSupportVectorMachine
MuhammadAgmalaro;ImasSitanggang
10:55 RecognitionofAcademicEmotionsinOnlineClasses TeeConnie
11:15ImageSteganographyCompressiveSensingOrthogonalMatchingPursuit
IrmaSafitri
ADVANCED PROGRAM | ICOICT 2021
49
Session 4C–DataScience
Venue RoomC:Arjuna
SessionChair FaizalMakhrus,S.Kom.,M.Sc.,Ph.D
Time PaperTitle Presenter
10:15IndonesianHoaxIdentificationonTweetsUsingDoc2Vec TitiWidaretna
10:35ElectronicNoseDatasetforClassifyingRiceQualityusingNeuralNetwork
FerdyErlangga
10:55SVMParallelConceptTestwithSMODecompositiononCancerMicroarrayDataset
RahmatPrasojoe
11:15DetectingOnlineRecruitmentFraudUsingMachineLearning HriditaTabassum
11:35DataminingforrevealingrelationshipbetweenGooglecommunitymobilityandmacro-economicindicators
GunawanGunawan
Session 4D–E-LearningandHCI
Venue RoomD:Nakula
SessionChair LewSookLing
Time PaperTitle Presenter
10:15SuitableKnowledgeManagementProcessImplementation:acasestudyofPTSinergiSentraDigital
YusufPratama
10:35CriticalSuccessFactorsforProjectTrackingSoftwareImplementation:ACaseStudyataBankingCompanyinIndonesia
HendroHadi
10:55 AssuranceCasePatternusingSACMNotation NungkiSelviandro
11:15SustainabilityAndAptnessOfGameElementsInAGamifiedLearningEnvironment
MageswaranSanmugam
11:35UserInterfaceModelforVisualizationofLearningMaterialsinComicStripFormUsingGoal-DirectedDesignMethod
DanangJunaedi
ADVANCED PROGRAM | ICOICT 2021
50
Session 4E–Networking,IoT,andSecurity
Venue RoomE:Sadewa
SessionChair ChongSiewChin
Time PaperTitle Presenter
10:15AccessibilityandResponseTimeAnalysisontheCOVID19WebsiteinIndonesia RyanWicaksono
10:35ModifiedBitParityTechniqueforErrorDetectionof8BitData
FakhiraZulfira
10:55IoTApplicationonAgriculturalAreaSurveillanceandRemote-controlledIrrigationSystems
RatnasariRohmah
11:15Present-80EncryptionAlgorithmImplementationonGPRSArduinoMega-2560CyberPhysicalTrackingSystem
RiniWisnuWardhani
11:35HuntingCyberThreatsintheEnterpriseUsingNetworkDefenseLog
ArdianOktadika
ADVANCED PROGRAM | ICOICT 2021
51
TECHNICALPARALLELSESSIONSCHEDULEParallelSession#5:Thursday,August5,2021,13:15–15:15UTC/GMT+7
Session 5A–Networking,IoT,andSecurity
Venue RoomA:Yudhistira
SessionChair OoiShihYin
Time PaperTitle Presenter
13:15BuildinganIDCardRepositorywithProgressiveWebApplicationtoMitigateFraudbasedontheTwelve-FactorAppmethodology
KevinAkbarAdhiguna
13:35 XB-Pot:RevealingHoneypot-basedAttacker'sBehaviors RyandyDjap
13:55DesignofaSnort-basedIDSontheRaspberryPi3ModelB+ApplyingTaZmenSnifferProtocolandLogAlertIntegrityAssurancewithSHA-3
RiniWisnuWardhani
14:15LearningMethodofPerformance-orientedCongestionControl(PCC)forVideoStreamingAnalysis
SidikPrabowo
14:35ExperimentalInvestigationofWaveAbsorberMadeofRingResonator-BasedAMCStructure IchsanNusobri
Session 5B–DataScience
Venue RoomB:Bima
SessionChair Dr.DyahArumingTyas,S.Si
Time PaperTitle Presenter
13:15 InformationCascadeMechanismandMeasurementofIndonesianFakeNews
AslaSonia
13:35FraudAccountsIdentificationModellingonMulti-PlatformE-Commerce
GrawasSugiharto
13:55ClassificationonParticipantsRenewalProcessinInsuranceCompany:CaseStudyPTXYZ
NoperidaDamanik;DeddyUtomo
14:15HybridSpace-TimeModelandMachineLearningforForecastingMultivariateSpatio-TemporalData HendriPrabowo
14:35ComparativeStudyofCovid-19TweetsSentimentClassificationMethods
UntariWisesty
ADVANCED PROGRAM | ICOICT 2021
52
Session 5C–DataScience
Venue RoomC:Arjuna
SessionChair MuhammadAlfianAmrizal,B.Eng.,MIS,Ph.D
Time PaperTitle Presenter
13:15CountDataForecastingusingPoissonAutoregression
forCOVID-19CasePredictioninJakarta BahrulNasution
13:35OptimizationofCropsAllocationPlanninginCianjurInvolvingWaterCostConstraintsUsingGenetic
AlgorithmBambangWahyudi
13:55 FakeNewsDetectionwithHybridCNN-LSTM KianLongTan
14:15AspectBasedSentimentAnalysisWithCombination
FeatureExtractionLDAandWord2vecRizkaVioOctrianyInggitSudiro
14:35SentimentAnalysisonBeautyProductReviewsusing
LSTMMethodMuhammadRafii
Danendra
Session 5D–DataScience
Venue RoomD:Nakula
SessionChair ChongLeeYing
Time PaperTitle Presenter
13:15IndonesianIDCardExtractorUsingOpticalCharacterRecognitionandNaturalLanguagePost-Processing
FirhanRusli
13:35AnalysisofRecordsManagementMaturityLevelforDataCollectionofNetworkAssetsinIndonesianTelecommunicationIndustry
RizkyEkaPutri
13:55DataAcquisitionGuideforForestFireRiskModellinginMalaysia
YeeJianChew
14:15 ImplementationofHiddenMarkovModel(HMM)toPredictFinancialMarketRegime
IrmaPalupi
14:35PredictionofGraduationwithNaïveBayesAlgorithmandPrincipalComponentAnalysis(PCA)onTimeSeriesData
WishnuHerlambang
ADVANCED PROGRAM | ICOICT 2021
53
Session 5E–DataScience
Venue RoomE:Sadewa
SessionChair OngThianSong
Time PaperTitle Presenter
13:15ComparativeAnalysisofSupportVectorMachine(SVM)andRandomForest(RF)ClassificationforCancerDetectionusingMicroarray
IrawansyahIrawansyah
13:35EvaluatingtheBPPTMedicalSpeechCorpusforAnASRMedicalRecordTranscriptionSystem
ElviraNurfadhilah
13:55ImplementationofSimulatedAnnealing-SupportVectorMachineonQSARStudyofIndenopyrazoleDerivativeasAnti-CancerAgent
IsmanKurniawan
14:15RansomwareDetectiononBitcoinTransactionsUsingArtificialNeuralNetworkMethods
HairilHairil
14:35EmotionalContextDetectiononConversationTextwithDeepLearningMethodUsingLongShort-TermMemoryandAttentionNetworks
AfridaHelen
ADVANCED PROGRAM | ICOICT 2021
54
ABSTRACTSESSION1A:ApplicationsforPost-PandemicRecovery
TheroleoftechnologycapabilitiesandinnovationcapabilitiesinachievingbusinessresilienceofMSMEs
duringCovid-19:EmpiricalStudy
GrisnaAnggadwita,RatihHendayani,ErniMartini,MuhammadKamil
Thisstudyaimstoidentifytheroleoftechnologicalcapabilitiesandinnovationcapabilitiesinthebusiness resilience of Micro, Small and Medium Enterprises (MSMEs) during the Covid-19pandemic.Thisstudyalsoexaminesthemediatingroleofinnovationcapabilityontherelationshipbetweentechnologycapabilityandbusinessresilience.Thisstudyusesaquantitativemethodwithacausalityapproachtoexaminetherelationshipbetweenvariables.Asurveywasconductedof400MSME owners in Jakarta, Indonesia using a random sampling technique. Structural equationmodels are used to predict and estimate relationships. The results of this study indicate thattechnological capabilities and innovation capabilities have a positive and significant effect onbusinessresilienceinMSMEsduringtheCovid-19pandemic.Inaddition,innovationcapabilityhasalsobeenshowntoplayasignificantroleasamediatorintherelationshipbetweentechnologicalcapability andbusiness resilience.This studyemphasizes that the increasing importanceof theconceptofbusinessresilience inthe faceof theCovid-19pandemicthusencouragingMSMEstoimprovetheirtechnologicalcapabilitiesandinnovationcapabilities.Gaze-ControlledDigitalSignageforPublicHealthEducationduringCovid-19Pandemic
SunuWibirama,SuatmiMurnani,IrawanDharmaSukawati,RidiFerdiana
Formorethanadecade,digitalsignageshavebeenusedinhealthfacilitiesandpublicenvironmenttoprovidefunandinteractiveapproachofeducation.Unfortunately,interactingwithconventionaldigitalsignageduringCovid-19pandemicraisesaconcernonitshygiene.Thus,touchlessinteractionis preferable to avoid direct contact on the touch screen. Here we present a novel study oneffectiveness of gaze-based interaction in a digital signage for public education about Covid-19.Insteadoftouchingthescreen,theusersengagewiththecontentbygazingatadynamicbuttonthatmovesinhorizontalorverticaldirection.Experimentalresultsshowthatgazingatfasterdynamicbuttons (angular speed of 60.28 ◦/s) than its slower counterpart (angular speed of 30.14 ◦/s)requiresshortertimetocompleteathree-stepstask(T=84,Z=−1.977,p<0.05).OurstudyprovidesascientificproofofconceptfordevelopmentoftouchlessdigitalsignagethatcomplieswithtechnicalguidelinesoftheWorldHealthOrganizationoncleaninganddisinfectingsurfacesinnon-healthcaresettings.
AnalysisoftheHouseofRisk(HOR)ModelforRiskMitigationoftheSupplyChainManagementProcess(CaseStudy:KPBSPangalenganBandung,Indonesia)
RatihHendayani,EllysaRahmadina,GrisnaAnggadwita,RinaPasaribu
ThisstudyidentifiesriskmanagementinthesupplychainofdairycompaniesinBandung,WestJava,Indonesia.Usingasinglecasestudymethodology,semi-structuredinterviewswereconductedwithvariousdepartmentsofsupplychainagentsworkingindairycompanies.Datawerecollectedaboutthedairyprocessingcompanysupplychainnetwork,manager'sknowledgeofsupplychainriskmanagement,andtherisksinherentinthemilksupplychain.Ingeneral,companieshavean
ADVANCED PROGRAM | ICOICT 2021
55
awarenessofsupplychainrisks.Byusing theHouseofRisk(HOR)model,whichcombines twomethods, namely Failure Mode and Effect Analysis (FMEA) and Quality Function Deployment(QFD), the risks identified in the supply chain have been assessed and prioritized. Severalrecommendationshavebeenmadetomitigatehighpriorityrisks.Thestudyfound15riskeventsinthecompany'ssupplychainprocess,11riskagentswheresevenofthembelongtothepriorityriskagentcategorywhichhasashareof79.55%ofthetotalriskexperiencedbythecompany.Todealwiththesepriorityriskagents,thecompanyfoundsevenriskmitigationactionstominimizetheimpactthatcouldharmthecompany.
EvaluationoftheSocialRestrictionanditsEffecttotheCOVID-19SpreadinIndonesia
InnaSyafarina,AyuShabrina,ArnidaLailatulLatifah,DiditAdytiaThegovernmentofIndonesiahasimplementedalarge-scalesocialrestrictionofvariouslevels,inthe local andnational region to control theCOVID-19 transmission over the country. Successfulsocial restrictions are believed as a powerful way in controlling COVID-19 spread. This paperevaluatedthesocialrestrictionthathasbeenimplementedinIndonesiaanditsconnectiontothespreadofCOVID-19.Thesocialrestrictionisquantifiedbythechangesinhumanmovement,whilethespreadofCOVID-19iscomputedbyagrowthrateanddoublingtimeoftheCOVID-19cases.Thispaper showed the social restriction reduced human mobility up to more than 50% when therestriction was started. It flattened the cases but it only lasted for approximately twomonths.Afterward,thegrowthrateandthedoublingtimeshowedanincreaseoftheCOVID-19casesanditwas getting much worse starting at the end of the year 2020. Moreover, we found that theimplemented social restriction in Indonesia is less effective to reduce the COVID-19 spread inIndonesiaashumanmobilityduringaholiday,anespeciallylongweekendismuchstronger.
ContributingClinicalAttributestoCOVID-19MortalityinJakarta:MachineLearningStudy
MuhamadErzaErzaAminanto,BahrulIlmiNasution,AndiSulasikin,YudhistiraNugraha,Juan
Kanggrawan,AlexLukmantoSuhermanSince December 2019, we have lived in a pandemic era of severe acute respiratory syndromecoronavirus2(SARS-CoV-2).MedicalrecordsofCOVID-19patientshavebeenreportedandanalyzedworldwide. TheHealthAgency of Jakarta, Indonesia, collected clinical symptoms, demographics,travelhistory,andmortalityinformationfromMarch2020uptonow.DespitemassiveresearchonCOVID-19patients'data,thesignificantclinicalsymptomsthatleadtoCOVID-19mortalityinJakartahavenotbeenwelldescribed to thebestof theauthors'knowledge.Weextracted theCOVID-19recordsinJakartaandcomparedthembetweenpatientswhoweredischargedanddeceased.Thispaperexamineseachclinicalsymptom'simportancetomortalityusingmachinelearningtechniques,namelyweightedArtificialNeuralNetwork,DecisionTree,andRandomForest.WeobservedthatPneumonia, Shortness of Breath,Malaise, Hypertension, Fever, and Runny Nose are the top sixsignificant clinical symptoms that lead todeaths in Jakarta.We suggestmedical expertsbecomemorecautiouswiththesesymptoms.Also,inmedicalfacilities,thesesymptomscanbeusedaspre-screeningbeforeenteringthefacilities.
ADVANCED PROGRAM | ICOICT 2021
56
ABSTRACTSESSION1B:ComputerVision
Multi-TargetRegressionUsingConvolutionalNeuralNetwork-RandomForestForEarly
EarthquakeWarningSystem
BenaldyYugaAdhaityar,DavidSahara,CecepPratama,AdiWibowo,LeniSophiaHeliani
Indonesiaoccupiesaveryactivetectoniczonebecausetheworld'sthreelargeplatesandnineothersmallerplatesmeeteachotherinIndonesianterritoryandformacomplexplatemeetingpath.EastJavaProvinceispartoftheSundanesearc,whichhasarelativelyhighlevelofseismicityandhasacomplexgeologicalsystemthatisaresultoftheIndo-Australianplate.Therefore,asystemthatcanprovideearthquakeearlywarning(EEW)isneededtoreducecasualties.Inthispaper,wedeterminethe location andmagnitude of the earthquake using the Convolutional Neural Network (CNN) asfeatureextractionandRandomForest(RF)formulti-targetregression.EarthquakesinEastJavain2009-2017areusedtotrainandvalidatetheproposedmodel.Basedontheexperiment,thelowesterrorobtainedfromtheCNN-RFmodelis18kmforlongitude,32.2kmforlongitude,and0.3490formagnitude.
Vision-BasedEmployeeActivityClassification
RizalKusumaPutra,EmaRachmawati,FebryantiSthevanieAnemployeeshouldbecompetentandexpertiseintheirrespectivefields.Anevaluationisneededtomaintainthequalityofemployee'sperformance,oneofwhichcanbedonebyobservingtheiractivityduringworkinghours.Thisresearchdiscussestheclassificationoftheemployee'sactivityindeskwork.Classificationofemployee'sactivityisinvestigatedusingResNetandtheCyclicalLearningRatemethodinanoveldataset,i.e.vision-basedemployeeactivity.Classificationisdonebylookingatthreetypesofemployeeactivities:talkingonthephone,usingaPC,andplayingsmartphone.Themostoptimalresultof this research is ResNet50 using CLRwith image input of 224x224x3,which has an accuracy of87.01%and12.99%errorratefortalkingonthephone,99.95%accuracyand0.05%errorrateforusingapc,81.67%accuracyand18.83%errorrateforplayingsmartphone,andhasadecreasinglossvalue. In addition, this research shows that cyclical learning rate significantly affects the modelperformance.
CompressiveSensingImageWatermarkingOrthogonalMatchingPursuit
IrmaSafitriEaseofaccesstodigitaldataontheinternetispronetorampantpiracy,especiallyofcopyrightandownership in themultimedia industry.Therefore,weneeda security system that canprotectandsecurethecopyrightofdataownership.Watermarkingisonesolutiontoovercomepiracythatoccurs.Watermarkingallowstheembeddedconfidentialinformationtobesolvedbyotherparties.Thisstudydiscussesimagewatermarkingusingthediscretewavelettransform(DWT)methodwhichisusedtodecomposethehostimageasanimagewherethewatermarkisembedded,thespreadspectrum(SS)methodusedintheinsertionprocessbyspreadingwatermarkbitsonthehostimageandcompressivesensing (CS)which is used to increase the efficiency of thewatermark technique to increase thecapacityandperceptibilityofthewatermarkingprocess.Thereconstructionprocesswascarriedoutusing orthogonal matching pursuit (OMP). The result is that the host image has a watermarkembedded in thebest sub-band.The test results showed theperformanceof thePSNRvaluewas53.3553dB,MSE0.30029,andBER0.20027.
ADVANCED PROGRAM | ICOICT 2021
57
TrafficSignRecognitionwithConvolutionalNeuralNetwork
ZhongBoNg,KianMingLim,ChinPooLee
ThispaperpresentsanablationanalysisofMultilayerPerceptronandConvolutionalNeuralNetworksin traffic sign recognition. The ablation analysis studies the effects of different architectures ofMultilayerPerceptronandConvolutionalNeuralNetworks,batchnormalisationanddropout.Atotalof8 different models are reviewed and their performance is studied. A traffic sign dataset withapproximately 5000 images is collected. The experimental results demonstrate that ConvolutionalNeuralNetworksoutperformMultilayerPerceptron ingeneral.Leveragingdropout layerandbatchnormalisationiseffectiveinimprovingthestabilityofthemodelthusyieldingbetterresults.
DeepConvolutionalGenerativeAdversarialNetworkApplicationinBatikPatternGenerator
AgusEkoMinarno,Moch.ChamdaniMustaqim,YufisAzhar,WahyuAndhykaKusuma
Batikhasbeenallegedasoneoftheoldestculturalheritagesworldwide.Sincetheinitiation,batikhasbeenidentifiedwithvarioustypesandpatterns.Variouskindsofbatikmakingtechniqueshavelongbeenpopularizedbythewidercommunity.DCGANservesasanewideaforthemodernbatikworld.Such algorithm is capable of reproducing a novel image to produce batik patterns previouslyunnotified.Hence,thisstudyaimstoproposeaDCGANmodeltocraftanewtypeofbatikpattern.ByutilizingalargedatasetofbatikimagesandavarietyoftheproposedDCGANmodels,theproposedalgorithmhassucceededincraftinganovelanddiversesyntheticbatikpattern.
ADVANCED PROGRAM | ICOICT 2021
58
ABSTRACTSESSION1C:DataScience
ConvolutionalNeuralNetworksforIndonesianAspect-BasedSentimentAnalysisTourism
Review
RoyanNayoan,AhmadFathanHidayatullah,DhomasHattaFudholi
Inrecentyears,electronicwordofmouth(e-WOM)hasbeenwidelyusedbypeoplearoundtheworld.Tripadvisorisane-WOMtravelwebsitewhichprovidesinformationaboutreviewsandopinionsontravel-relatedcontent.Tohelpusersgather information faster, aspect-basedsentimentanalysis isnecessary.Aspect-based sentiment analysis helpsusers to capture and extract important featuresfromthereviews.Therefore, thisstudyaimstobuildanaspect-basedsentimentanalysismodelofIndonesian tourism review by extracting aspect-category and their corresponding polarities fromuser reviews. To gain the bestmodel, we performed several experiments by using ConvolutionalNeuralNetworks(CNN).Moreover,wecomparedourCNNmodelwithCNN-LSTMandCNN-GRUtoidentify thesentimentandaspects fromthereviews.Wealsoperformednegationhandling inourfeatureextractionprocess to improveourCNNmodels.Basedonourexperiments,CNNcombinedwith both POS tag and negation handling outperformed the other models with the accuracy ofsentimentanalysisof0.9522andaspectcategoryof0.9551.
SentimentAnalysisonMarketplaceReviewusingHybridLexiconandSVMMethod
MuhammadMukhtar,WikkyFawwazAlMaki,AdeRomadhonyNowadays,especiallyduringtheCovid-19pandemictime,thereisenormousriseinonlinetransactions.Thereareseveralpopularmarketplacesthatprovidereviewfacilitytohelpcustomerschoosingtherightproductsorsellers.SentimentanalysisisastudytoclassifyareviewtexttosentimentclassesTables.Inthispaper,wepresentastudyofsentimentanalysisonmarketplacereviewtextusinghybridmethod: based on lexicon fromSentiwordnet 3.0 and SupportVectorMachine (SVM)method. Theexperimental results show that the hybrid method outperforms the lexicon approach and SVMapproach.
ForecastingNumberofCOVID-19CasesinIndonesiawithARIMAandARIMAXModels
BimoSatrioAji,Indwiarti,AniqAtiqi
DuringthepandemicCOVID-19,Indonesiahasasignificantnumberofpositivecasesamongcountriesin Asia. In early December 2020, the death rate in Indonesia had been reached more than 3%.Meanwhile, the daily number of positive is also continued to increase, it happens due to lack ofanticipationrulesmadebylocalauthoritiesandcentralgovernment.Thus,thepreventivestepsuchforecastingbecomesamajorissueintheareaofscienceandtechnology,tomakeallstakeholderswell-prepared against this pandemic. This paper provides the performance of The AutoregressiveIntegratedMovingAverage(ARIMA)toforecastseveralCOVID-19andalsoexaminesAutoRegressiveIntegratedMovingAveragewithexogenousvariables(ARIMAX)modelbyconsideringGoogleTrendsasanexternalvariable.WeconsideradailydatasetfromtheofficialwebsiteoftheJakarta'sCOVID-19andtheGoogleTrendsdatabasedoncertainqueriesasexternalvariablesonMarch1-November25, 2020. According toARIMA andARIMAXmodels,we haveARIMAXmodelwithGoogle TrendsimprovingARIMA'sperformancebyreducingtheMAPEby0.8%.
ADVANCED PROGRAM | ICOICT 2021
59
DisasterTweetClassificationBasedOnGeospatialDataUsingtheBERT-MLPMethod
IqbalMaulanaandWarihMaharaniAsapopularsocialmediaintheworldandeveninIndonesia,Twitterhasavarietyofpopulartopicsmakingthesetopicstrending,includingthetopicofnaturaldisastersthathaveoccurredinIndonesia.TheDKIJakartaflooddisasterinearly2020madeabigsceneontrendingtwittertopics.Thisstudyaims to classify these tweets into "flooded" and "not flooded" predictions with the tweets andgeospatial features. The model proposed for classifying is BERT-MLP. Bidirectional Encoder fromTransformers (BERT) is used in the pre-trained model to classify these tweets and Multi LayerPerceptron(MLP)isusedtoclassifygeospatialfeatures.Thescenariodesignedforthemodelfocusesonthepreprocessingoftweetsasfollowswithoutstopwordremoval,withoutstemming,withboth,andwithoutboth.Onceclassified,thetweetwillbevisualizedintoatwo-dimensionalinteractivemap.Thebestscenarioresultshaveanaccuracyof82%inscenarioswithoutstemmingandwithstopwordremoval.Thisisduetothestemmingprocesseliminatessomeofthefeaturesintweetsaround6%.Thisstudyalsoshowstherelationshipbetweentheinfluenceofnegativecontexttweetsonthe"notflooded"classwithanorientationof65%ofthetotaldata.However,definingmanualstopwordscanaffectbecausestopwordremovalwillnotdeletewordsthatstillhavecontextrelatedfeaturestothetopic.
CyberbullyingDetectiononIndonesianTwitterusingDoc2VecandConvolutionalNeuralNetwork
ShindyTrimariaLaxmi,RitaRismala,HaniNurrahmi
Cyberbullyingistheactofthreateningorendangeringothersbypostingtextorimagesthathumiliateorharasspeoplethroughthe internetorothercommunicationdevices.Accordingtoasurvey fromPollingIndonesiaandAsosiasiPenyelenggaraJasaInternetIndonesia(APJII)aboutcyberbullying,49%of 5900 participants claimed they have been bullied. Therefore, this research conducted with theintention topreventcyberbullyingacts,especially in Indonesia.Wecollected thedata fromTwitterbasedonTwitter'sTrendingkeywordswhichcorrelatedtocyberbullyevents.Thenwecombineditwiththedatafrompreviousresearch.Weobtainedatotalof1425tweets,consistsof393datalabeledascyberbullyand1032datalabeledasnon-cyberbully.Thereupon,webuildtheDoc2Vecmodelforfeaturesextraction,andaclassifiermodelusingthebaselineclassificationmethod(SVMandRF)andCNNtodetectthecyberbullytexts.TheresultsshowsthattheclassifierusingCNNandDoc2vechasthehighestF1-score,65.08%.
ADVANCED PROGRAM | ICOICT 2021
60
ABSTRACTSESSION1D:E-LearningandHCI
DigitalNudgeEvaluationonCOVID-19tracingApplication
DyahWahyuSukmaningsih
Nudgeisconsideredasaninterventiontochangeuserbehaviorandinfluencedecisionmaking.Mobileappshavebecomeapartofoureverydaylife.Inthispandemicera,governmentsusemobileapps'technologytocontrolthespreadofCOVID-19infection.ManygovernmentsimplementedCOVID-19tracingapps,andcitizensareforcedtousetheappstorecordtheircontacttracingandalertthemiftheyhavecontactwithaninfectedperson.Thekeytocontroltheincreasingspreadofthevirusalsodepends on citizens adhere to health protocol. Government experience difficulty in enforcing theprotocol without strict surveillance. Therefore, building awareness of risk is essential-this studyattempttodesignnudgeinterventionintoCOVID-19tracingapps.Fromtheevaluationoftheoriginalapps,fournudgeinterventionscanbeappliedintheapplication.DevelopingSuicideRiskIdeaIdentificationforTeenager(SERIINA)MobileAppsPrototype
usingExtendedRapidApplicationDevelopment
TeniaWahyuningrum
Suicidedeathisthenumber15causeofdeathintheworld.Suicidecasesgoundetectedbecausetheperpetrator showsno signs beforehand. Therefore, it is necessary to identify the risk of suicide inadolescentsearly,accessedquickly,maintainsuserprivacy,andunderstanduserneeds.UsingtheRiskFactors of Suicidal Ideation (RFSI) questionnaire, it hoped to detect early suicide incidents inadolescents.ThisresearchproposedaSuicideRiskIdeaIdentificationforTeenager(SERIINA),amobileapplicationdevelopedusingtheExtendedRapidApplicationDevelopment(ERAD)method.TheERADmethodcombinestheconceptofdesignsprintandRADinthesystemdevelopmentcycle.Basedontheresearchresults,SERIINAmobileappsdevelopment iscompletedin19days.Thefunctionaltestingusing Black-Box testing shows that the application works well and compatible in the five mainscenarios.Theresultsofinterfacedesigntestingusingheuristicevaluationindicatethattheapplicationhasbeenwelldesigned,withausabilityvalueof2.34orabout72-85%accordingtodesignrules.
DesigningAnEducationalGameEvaluationFrameworkBasedOnGameMechanic
SatrioARukmono,FaisZharfanAzif,MuhammadZuhriCaturCandra
Childrenineverydaylifeareincreasinglyusingeducationalgames.However,thequalityofeachofthemanyeducationalgamesavailablevaries.Someevaluationframeworksexist,butmostarepronetotheevaluator'ssubjectivity,whichcannotbecomparedobjectively.Thisstudyaimsto formulateaframeworkthatevaluatesthequalityofeducationalgamesobjectivelybasedonthegamemechanicsused.TheframeworkisbuiltuponBloom'staxonomyasthebasistoascertaintheacademicsideandMDA (Mechanics-Dynamics-Aesthetics) Framework todistinguish the game side.Then, it assesseseacheducationalmechanicbasedonastandardintheevaluationframeworktoobtainanaccurate,quantifiablescoreasameasure.Validationoftheframeworkinvolvesusingtheframeworktoevaluateexistingeducationalgamesandcomparingtheresultswithexpertreviews.Withthisframework,aneducationalgamequalitycanbemeasuredobjectivelyandquantitativelybasedonthetechnicalandfundamentalelementsthatexistineachgame.
ADVANCED PROGRAM | ICOICT 2021
61
CultivatingRecyclingAwarenessinPreschoolersusingAnimatedInteractiveComic
SitiZulaihaAhmad,AriffinAbdulMutalib,NurulAinaKamarulzamanRecyclingawareness is a crucial social responsibility thatneeds tobe cultivated in the communityespecially to the young generation. Knowledge of recycling awareness is less exposed to children,especiallyinschoolsinceitisanactivitythatisnotappliedbyparentsathomebecausetheyarebusywith work. Therefore, this study introduces an interactive animated comic of recycling, as analternative learning application to facilitate children in preschool in learning and gaining newknowledge. I-Recycle consists of numerous storylines such asweekly activities, collecting recycleditemsathomewithfamily,cleaningtheparkwithfriendsandlearningactivitiesintheclassroom.Thisstudy integrates all components using three-phase development model, which consists of pre-development,development,andpost-developmentphases.Usabilitytestinghasbeenconductedwith20 children in individual setting to ensure that i-Recycle is usable and learnable, as those are theimportantelementstocultivateawarenessamongthechildren.Theresultrevealedthat79.75%oftheparticipantsagreedi-Recycleisapplicable,easytouseandeasytolearn.Infact,theuseracceptanceresults also proven that the application is useful, easy to use and acceptable (μ = 3.9). The resultdisclosedthati-Recycleisinterestingfortargetuserstogainnewknowledgeandlearnaboutrecyclingthroughstorytellingandanimatedcomic.However, it requires improvement in termsofaudioandinterfacedesigntoensuretheapplicationcouldsufficetheneedsofpreschoolers.
AnalysisInfluenceofTheOrganizationalLearningEnvironmentFactorsToEncourageEmployeeMotivationUsingE-Learning
NanikQodarsih,AchmadHidayanto,MuhammadRifkiShihab
Thesuccessfuluseofe-learningofcoursedependsonhowmuchgreatemployeeshaveengagementtoe-learning.Relatedtothiscase,organizationsthatrelyontechnologye-learningsoweneedtobuildmotivationofemployees.Wecanbuildemployeemotivationwhenusee-learning,onethingthatthewriterlookimportantisorganizationalsupportlearningenvironment.Thisstudywillanswerhowtheinfluence of the organizational learning environment to encourage employee motivation using e-learning.TheresearchwasconductedinIndonesiaJudicialTrainingCentre.Theresearchmethodologyusedinthisstudyisacasestudyresearch,researchsurveyandquantitativeresearch.Thenumberofrespondentswho tookpart in the surveyamounted to318 responses.Thequestionnairedatawasprocessedusing theCBSEMmethod.Dataprocessing isdoneusing theAMOS22application.Thisresearchhasbeenproven toprovideanewperspectiveon the fieldof technologyand informationsystems how the influence of organizational learning environment factors such as managementsupport,technologysupport,organizationalsupport,friendsupport,andjobsupportcanaffecttheSelfDeterminationTheoryandTechnologyAcceptanceModelvariablesine-learningacceptance.
ADVANCED PROGRAM | ICOICT 2021
62
ABSTRACTSESSION1E:Healthcare,Bioinformatics,and
BiomedicalApplication
DistributedPhylogeneticTreeProcessingonBiologySequencesUsingMapreduce
RenaningSusilo,Setyorini,SitiAmatullahKarimah
MultipleSequenceAlignment(MSA)isanimportantprocessintheanalysisofbiologicalsequencesbymakingcomparisonsofseveralbiologicalsequences.InsomeMSAalgorithms(suchasCLUSTALW,for example), the formation of phylogenetic trees as a guideline in the alignment process has animportant role in determining the accuracy of the final alignment results. From the whole MSAprocess,theformationofphylogenetictreecomputationaltimeincreasedasthenumberofsequencesincreases.Computing thesimilarity score forall sequencepairs takesa lotof time, causinga longcomputationtimeproblem.Inaddition,thisresearchusestheneighborjoinmethodasatechniquetobuild a pyloghenetic tree This Research examines the potential efficiency of computationalphylogenetictreesinparallelandisdistributedtotheHadoopenvironmentusingMapReduce.TheresultsshowedthataphylogenetictreecanbegeneratedusingMapReducecomputationandresults19%timeefficiencyfrom2and3computationnode.
LinearRegressionModeltoPredicttheSpreadofCOVID-19inTangerangCity
YusufSudiyono,AgungTrisetyarso,HarjantoPrabowo,MeylianaMeyliana
TheoutbreakofacuterespiratorysyndromevirusdiseaseinChinaattheendof2019hascausedaglobalepidemicaswellashighmortalityratesinaffectedcountries.ThisresearchaimedatexaminingtheextentofthespreadofconfirmedCovid-19casesinTangerangCity.ThedatausedincludedthedataofconfirmedCovid-19patients.Suchdatawasintegratedwithgeospatialdatafoundin13sub-districtsinTangerangCity.ThepredictionofthespreadofconfirmedCovid-19caseswasmadebyusingLinearRegressionmodel.TheresultsoftheMAPEcalculationwithavaluebelow10%in13districtsresultedinaverygoodpredictivevalue.Thispredictionresultedinagraphandwasconnectedtoeachotherinathematicmapcoordinatepointsystem.TheresultsoftheCovid-19spreadpredictionweredividedintoseveraldistrictsandindicatedwithdifferentcolorvariations.Therefore,thedarkertheresultingcoloronthethematicmapvisualization,indicateanincreaseinCovid-19casesthathaveoccurred.StrategicInformationSystemPlanningforIndonesiaNon-franchisePharmaciesBasedonJohn
WardandFactorAnalysisMethod
TabahArwiyanto,AdianFatchurRochim,RRizalIsnanto
ThisstudyaimstoproduceastrategicplanningproposalfortheIS/ITofnon-franchisepharmaciestobeabletocompetewithfranchisedpharmaciesandtorealizethevision,mission,andobjectivesofthe pharmacy. This researchwas conducted using amixed-methodwith a sequential explanatorydesign.Observationandinterviewmethodsareusedtoformulatethevision,mission,andobjectivesofthepharmacy.ThedataobtainedisusedasabasisformakingobjectivefactorsintheformofLikertscale questions. The results of the questionnaire were analyzed for validity and reliability usingPearson'sProductMomentandCronbach'sAlphaandshowedtheresultsofther-countvalueofallthequestionswerepositiveandgreaterthanther-table(r-table>0.457)withavalueofα=0.980,whichmeans thequestion items in thequestionnairevalidand reliable.Thesequestionsare thendescribedintheformofadecompositiondiagramwiththeconsiderationofvaluechainanalysis.Theoutput of the decomposition diagram is used as data for analyzing the company's success factors
ADVANCED PROGRAM | ICOICT 2021
63
(CSFs).QuantitativefactoranalysiswasalsocarriedoutwiththehelpofstatisticalapplicationsusingBartlett'sTestofSphericityandtheKeizer-Meyer-Olkin(KMO)test.Asaresult,thechi-squarevalueis279.591andtheMSAvalueis0.738,soitcanbeconcludedthatthedatahasmettherequirementsforfactoring.Basedonthefactoranalysis,7(seven)componentswerealsoextractedwhichwerethenused to compile the application portfolio in the McFarlan quadrant. The Customer RelationshipManager(CRM)applicationandthePointofSales(POS)applicationoccupyastrategicposition,whiletheInventoryModule,FinanceModule,SalesModuleoccupyahighpotentialposition.FlexibleMulti-LayerConduraFabricUltraWide-BandAntennaForTelemedicineApplication
YusnitaRahayuandTasyaKirana
Inthemedicalworld,acheck-upisthefirststepdoctorstaketoobtaininformationaboutapatient'shealth.Many advances in telemedicine, includinghealth tracking, have advancedwith the times.Awearableantennawasdevelopedtoenhancethemonitoringprocessandmakeitmoreconvenientforpatients.Fabricsthatarebothflexibleandcomfortableusedareoftenapplied.Itisalsonecessarytohaveanantennawithsufficientdatatransmittingcapacity.AnUltraWide-Band(UWB)antennaisadevicethatcantransmitlargequantitiesofdata.ThisisbecausetheUWBantenna'sbandwidthhasawidetransmissionband.ThispaperproposesaUWBflexibletelemedicineantennaworkingat6.55GHz. The condura substrate is used with frog patch shaped. Condura fabric was proved to be aninteresting fabric for textile antennas because of its strength, constant thickness, and high waterresistancepropertiesFromsimulatedresults,theantennahas7.86GHzofbandwidthwith10.8dBiofgain.Thesimulatedreturnlossof-20.57dBisachievedat6.5GHz.
AnonymizingPrescriptionDataAgainstIndividualPrivacyBreachinHealthcareDatabase
DediGunawan,FatahAlIrsyadi,YusufSulistyoNugroho,MaryamM
Prescriptiondataisasubsetofthehealth-relateddatawhichcanbecollectedbydrugstoreduringthepatient'smedicationperiod.Ingeneral,prescriptiondataconsistsofasetoftransactionrecordswhichcontains patients name or patients identification number and their prescribed medicine name.Analyzingsuchdatausingdataminingtechniquesbringsvariousadvantagesfordrugstores.However,performingdataminingtaskisnottrivialforthedrugstoresandpossiblythedrugstoredispatchestheprescription data to another party for data analysis.While it can solve the data analysis problem,unfortunately, such activity may result in privacy breach since sensitive information i.e., types ofpatients'diseasedue to thedataminerhasbackgroundknowledge to infercertainmedicine to thediseasetype.Toguaranteeindividualprivacypreservingofthesensitiveinformationamethodcalleddataanonymizationshouldbeemployedpriortohandlingtheprescriptiondatatoanotherpartyfordata mining purpose. Current data anonymization technique such as suppression technique cansuccessfullyaddresstheproblem,however,itresultsinsignificantitemlostandconsequentlyalotofuseful information is lost from the database. Tominimize the side effect of the suppression basedtechnique, a data anonymization which is based on swapping techniques can be a solution.Experimental results show that the swappingmethod successfullyprotects individualprivacywithrespecttoreducethenumberofitemlost.
ADVANCED PROGRAM | ICOICT 2021
64
ABSTRACTSESSION2A:Networking,IoT,andSecurity
SimulationOfJellyfishTopologyLinkFailureHandlingUsingFloyd-WarshallandJohnson
AlgorithminSoftwareDefinedNetworkArchitecture
MuhammadAriefNugrohoandAndrianRakhmatsyah
DatacenterdesignedtoaccommodateallITinfrastructure,whichiscurrentlyingreatdemandbythecompanies. In the data center, there are several types of network topologies, one ofwhich is theJellyfish Topology. The main problem in Jellyfish is link failure. Link failure causing connectivitybetweendevicesgoesdown. Inorder to solve thatproblem, it isnecessary todesignandsimulatebackup path for the networks. The emerging technology in networks called Software DefinedNetworks(SDN).UsingSDN,administratorcaneasilyprogramthenetworkviacontroller.SDNcansolvetheproblembyconfiguringthecontrollerinordertocreatebackuppathon-the-fly.Inordertofind the optimum backup path when link failure occurs in Jellyfish topology, floyd-warshall andjohnsonalgorithmareusedinthispaper.Simulationsarecarriedoutwithparametersconvergencetime, packet loss and throughput. Based on our experiments, FloydWarshall outperform Johnsonalgorithminallparameters.
GameTheoreticalPowerControlinHeterogeneousNetwork
AnggunFitrianIsnawatiandMasAlyAfandiThedevelopmentofwirelesstechnologyhaspenetratedinfemtocellcommunicationsystems,wherethis communication system is very flexible to develop. However, with many networks runningsimultaneously,which iscalledaheterogeneousnetwork,acombinationofmacrocelland femtocellnetworks,interferencebetweennetworksisunavoidable.Toresolvethisinterference,itcanbedonewithadaptivepowercontroltechniquesbytheuser.OneoftheadaptivepowercontrolmethodsisGameTheory.TheuseofGameTheoryonpowercontrolorGameTheoreticalPowerControlisoftenreferredtoPowerControlGame(PCG).Bydeterminingtheappropriateutilityfunction,theoptimalpowerisobtainedwhenusingthepowerupdateiterationprocess.TheresultsshowthatintheProposedmethodwhenitreachesaconvergentcondition,bothfemtousersandmacrousersareabletoreachSINRthatexceedsthetargetSINRof5.496forfemtousersand10.04formacrousers.Meanwhile,theDistributedPowerControl(DPC)methodisonlyabletoachievetheSINRuservaluewhichisthesameasthetargetSINR,whichis5and10forfemtousersandmacrousers,respectively.TheProposedmethodproducesahigherSINRvaluefortheuserthantheDPCmethodsothatintermsofachievingthetargetSINR,itcanbeconcludedthattheProposedmethodisbetterthantheDPC.
IoTDroneCameraforaPaddyCropHealthDetectorwithRGBComparison
ElvarettaYucky,AjiGautamaPutrada,MamanAbdurohmanThispaperproposesthesystemofpaddycrophealthdetectorusingdronecamera.Indonesiaisanagriculturalcountrythathasverylargeagriculturalland,whereeveryplanthealthmonitoringactivityisdonemanually.However, applying technologicaldevelopments in landmonitoringactivitieswillshortentimeandincreaseworkefficiency.Inthispaper,dronewitharaspberrypicamerahasusedtocaptureseveral imagesof rice fields fromseveral regions.The imagedatawillbeprocessed intoadigitalleafcolorchart(LCC)throughtheprocessofimageacquisition,RGBcolorextraction,andk-NNclassification.ThedatahasbeencomparedwiththerealLCC,whichisareferencetothehealthcolorofriceplants.Thepaddyfieldsthatareusedastheresearchmaterialare25daysafterplanting.The
ADVANCED PROGRAM | ICOICT 2021
65
resultshowsthattheprecisionofthemethodis88.89%,therecallis93.02%,theaccuracyis98.22%,andthespecificityis98.77%.
AReviewonIoTwithBigDataAnalytics
AbuFuadAhmad,Md.ShohelSayeed,ChooPengTan,KimGeokTan,MdAhsanulBari,FerdousHossain
TheInternetofThings(IoT)isapowerfulandtransformativeforcefortheconvergenceofthephysicalanddigitalworldoftechnology.TheIoTisconnectingthings,businessesandpeopleinreal-timeandonamassive scale. The IoT is actually the network of interconnected devices that contains actuators,sensors,electronics,softwareandconnectivitywhichletsthesethingsconnect, interactandtransferdata.ConnecteddevicesandsoftwareworkinwaysthatproducemassiveamountsofdatawhereBigDatacomes intothepicture.Bigdata is thediversesetsof informationthatarebothvery largeandcomplexinnature.Bigdataoffersabetterwayofmanagingandusingalargeamountofdatawiththeopportunitytoconductdeeperandricheranalysis.AlthoughtheextensivenumberofworksdoneonbigdataanalyticsandIoT,theoverlappingofthesetwofieldsofstudycreatesvariouspossibilitiesforthrivingdataanalysis in the IoTenvironment.Thisarticleprovidesa thoroughreviewof therecentadvancementofIoTwithbigdataandanalytics.Wealsomakeareviewoftherelationshipbetweenthese fields. This article discusses the application area of IoT and big data analytics aswell as theopportunitiescreatedbyenablinganalyticsinanIoTsystem.
VehicleBlindSpotAreaDetectionUsingBluetoothLowEnergyandMultilateration
MuhammadRezaWidyaPratama,MamanAbdurohman,AjiGautamaPutradaBlindspotistheareaaroundthevehiclethatcannotbeseenbythedriver'sviewevenwiththehelpoftherearviewmirrorsonthevehicle.Thereforeasolutionisneededtoovercomethisproblem.Thisstudyproposesasystemthatcandetectotherroadusersintheblindspotarea.ThissystemusestheBluetoothLowEnergy(BLE)moduletodetectthepresenceofobjects(smartphones)belongingtootherroadusers.PredictionofthedistancetotheroaduserisobtainedbycalculatingtheRSSIvaluebetweentheuser'ssmartphoneandthedetectingBLEmodule.Meanwhile,theobject'spositionisobtainedbyusingamultilaterationalgorithm. Inthisstudy, theKalmanFilter isalsousedtosuppressthenoiseobtainedduringthetargetdetectionprocess.Detectionresultsintheformofuserdistanceandlocationpredictionaredisplayedonthedriver'ssmartphonescreen.Testingthedetectionofvehiclesinablindspotareainstationaryconditionsproducesanaccuracyvalueof88.3%.
ADVANCED PROGRAM | ICOICT 2021
66
ABSTRACTSESSION2B:ComputerVision
FacialEmotionRecognitionusingTransferLearningofAlexnet
SarmelaRajaSekaran,ChinPooLee,KianMingLim
ThispaperpresentsadeeplearningmethodbasedonAlexNetarchitectureforemotionrecognition.Theideabehindourproposedframeworkistransferlearning.Wemanipulatecertainlayersofthepre-trainedAlexNetmodelandfine-tunethemodelonthepubliclyavailablefacialexpressiondatasetssothatitcanperformemotionrecognition.Theproposedmodelistrainedandtestedontwowidelyusedfacialexpressiondatasets,namelyextendedCohn-Kanade(CK+)datasetandFERdataset.Theproposedframeworkoutperformstheexistingdeeplearningmethodsinfacialemotionrecognition.
VisuallySimilarHandwrittenChineseCharacterRecognitionwithConvolutionalNeuralNetwork
WeiHanLiu,KianMingLim,ChinPooLee
Computervisionhaspenetratedmanydomains, for instance, security, sports,healthandmedicine,agriculture, transportation, manufacturing, retail, and so like. One of the computer vision tasks ischaracter recognition. In this work, a visually similar handwritten Chinese character dataset iscollected.Subsequently,anenhancedconvolutionalneuralnetworkisproposedfortherecognitionofvisuallysimilarhandwrittenChinesecharacters.Theconvolutionalneuralnetworkisenhancedbythedropoutregularizationandearlystoppingmechanismtoreducetheoverfittingproblem.TheAdamoptimizerisalsoleveragedtoaccelerateandoptimizethetrainingprocessoftheconvolutionalneuralnetwork.Theempiricalresultsdemonstratethattheenhancedconvolutionalneuralnetworkachievesa97%accuracy,thuscorroborateithasbetterdiscriminatingpowerinvisuallysimilarhandwrittenChinesecharacterrecognition.
PneumoniaClassificationusingGabor-ConvolutionalNeuralNetworksandImageEnhancement
AgusEkoMinarnoandMuhammadAlfarizy
Pneumoniaisarespiratorydiseasecausedbybacterialand,viralorfungalinfectionsandhasahighmortalityrate.Pneumoniaisusuallycharacterizedbythepresenceoffluidintheairsacsofthelungsoralveoli.IdentificationofpneumoniacanbedonewithChestX-Rayimage,buthamperedbyotherlungproblems thathavebeenexperiencedby thepatient.Therefore, in this studyproposedDeepLearningwith CNNmethod to solve classification problems quickly and precisely. In solving thisproblem, this research proposed Gabor Filter-Convolutional Neural Network method and ImageEnhancementpre-processingtechnique.Beforetheimageisprocessed,dataaugmentationwillalsobecarriedoutbyseveral techniques.TheuseofGaborFiltermanaged togetgoodaccuracywhileImageEnhancementwaslesssuitableforuseinthiscase.However,combiningimageenhancementandGaborfiltergetssmallerloss.ThisstudyobtainedthehighestaccuracywiththeGaborFilter-CNNmodelof94.9%andgetlowestlosswiththeImageEnhancement-CNNmodel35.8%.
ADVANCED PROGRAM | ICOICT 2021
67
FingerprintEnhancementusingIterativeContextualFilteringforFingerprintMatching
BramaYogaSatria,AgusBejo,RisanuriHidayatFingerprintmatchingdependsonthequalityofthefingerprintimages.Whenfingerprintimagequalityis low, itcandegradetheperformanceof fingerprintmatchingsignificantly.Fingerprint imagesareoftencontaminatedbynoise.Therefore,imagequalityiscrucialforfingerprintmatching.Inthispaper,an imageenhancementalgorithminwhichcontextual filtering isapplied iterativelytoa fingerprintimagehasbeenproposed.ThemainideaofthealgorithmistoiteratetheoutputoftheGaborfiltertogetbetterenhancementandmatchingperformance.Theresultofthealgorithmhasfivefilteredimagesduetofivetimesiteration.ItshowedthattheproposedmethodissignificantlybetterbasedonEqualErrorRate(EER)comparedtotheGaborfilterandthemodifiedGaborfilter.TheproposedmethodsurpassedtheGaborfilterby3.08%andthemodifiedGaborfilterby2.95%.
HistogramofOrientedGradientRandomTemplateProtectionforFaceVerification
LucasChongWeiJieandSiewChinChongPrivacypreservingschemeforfaceverificationisabiometricsystemwhichisembeddedwithtemplateprotectiontoprotectthedatainensuringdataintegrity.Inthispaper,anewmethoddubbedHistogramof Oriented Gradient Random Template Protection (HOGRTP) is proposed. The proposed methodutilizesHistogramofOrientedGradient approach as a feature extraction technique and fuseswithRandom Template Protectionmethod. The proposedmethod acts as amulti-factor authenticationtechniqueandaddsalayerofdataprotectiontoavoidthebiometriccompromisingissueduetothefactthatbiometricisirreplaceable.TheperformanceaccuracyofHOGRTPistestedontheunconstrainedface imagesusing thebenchmarkeddataset,LabeledFace in theWild (LFW).Apromisingresult isobtainedtoprovethatHOGRTPachieveshigherverificationrateinpercentagethanthepurebiometricscheme.
ADVANCED PROGRAM | ICOICT 2021
68
ABSTRACTSESSION2C:DataScience
Aspect-BasedSentimentAnalysisinBeautyProductReviewsUsingTF-IDFandSVMAlgorithm
NadiraPutriArthamevia,Adiwijaya,MahendraDwifebriPurbolaksono
Productreviewsareessentialine-commerceastheycanhelppotentialbuyersmakedecisionspriortomakingpurchasesandhelpsellersgetthemeasureoftheirproducts.Aproductcanhavethousandsofreviews,makingitburdensomeforpotentialbuyersandsellerstodrawaconclusionfromthoseabundantreviews.ThisresearchbuiltasystemthatappliesAspect-basedSentimentAnalysis(ABSA)withadatasetfromproductreviewsontheFemaleDailywebsite.ThesystemwasbuiltusingTF-IDFasitsfeatureextractionmethodcombinedwithwordbigramandwordbigram.TheSupportVectorMachine(SVM)algorithmisusedtoclassifythesentiments.Thisexperimentresultsindicatethatthepreprocessingstage,especiallythestemmingandstopwordsremovalprocessaregreatlyaffecttheaccuracyresults.ThechoiceofwordN-gramisalsocrucial,wherethisresearchshowsthatthewordunigramgivesahigheraccuracythanthewordbigram.ThefinalresultsofthisresearchshowthatTF-IDFcombinedwithwordunigramandSVMwithalinearkernelbringsoutthebestaccuracy,thatistosay,88.35%.
AspectTermExtractionUsingDeepLearning-BasedApproachonIndonesianRestaurantReview
RachmansyahAdhiWidhianto,AdeRomadhony
Aspect term extraction is a fundamental process in aspect-based sentiment analysis. Aspect termextractionaimstoidentifythereviewtextspanthatcontainstheaspectmentions.Inthispaper,wepresentourworkonaspecttermextractionforIndonesianrestaurantreviews,usingadeeplearning-basedapproach.WecollectedandannotatedanIndonesianrestaurantreviewsdataset,obtainedfroma restaurant reviewwebsite.Weperformed theannotationata token-levelandused the followingaspectlabelstoannotatethereviews:FOOD,PRICE,AMBIENCE,SERVICE,andMISCELLANEOUS.Thispaper treats aspect extraction as a token-level classification.We employed a ConvolutionalNeuralNetwork (CNN) model and Long Short-Term Memory (LSTM) model for the classification. TheexperimentalresultshowedthattheLSTMmethodgivesthebestperformance,withthemicroaverageF1-scoreis55,1%.
SpamDetectiononIndonesianBeautyProductReview
MuhammadAhsanAthallahandAdeRomadhony
Aproductreviewisoneofthemostimportantsourcesofinformationwhichcanhelpconsumerstofind themost suitable products for their needs. However, there is a chance a reviewer has otherintentionsthanprovidinganhonestreview,includingtoadvertisethebrandorotherbrands.Areviewthatdoesnotcontainanyinformationrelatedtotheproduct'saspects/featurescouldbeconsideredspam.Thispaperpresentsourworkonspamreviewdetection,specificallyinthedomainofbeautyproducts.We used SVM and Logistic Regression classifier and the following features: the reviewsentiments, product-related features, and review-centric features extracted from the reviews.Weclassifiedthebeautyproductreviewtextsasspamandnon-spamreviews.Theexperimentalresultshowed that the best accuracy percentagewas 81%, obtainedwhenwe used the sentiments andreview-centricfeatureswiththeSVMalgorithm.
ADVANCED PROGRAM | ICOICT 2021
69
EmotionClassificationonIndonesianTwitterUsingConvolutionalNeuralNetwork(CNN)
FirhanMaulanaRusli,RitaRismala,HaniNurrahmiHumansareinseparablefromemotions,emotionsfillhumanlifeatalltimes.Emotionshaveanimpactonsocialrelationships,memoryanddecision-making.Intheeraofthisresearch,humanstendedtoexpressemotionsthroughsocialmediasuchasTwitterintheformofvideos,imagesandtext.Overtime,socialmediahasbecomeanimportantpartofmostpeople'slives.Emotiondetectionisaresearchareathatiswidelyresearched,especiallyinthefieldoflinguistics.Therefore,weclassifyemotionswithaConvolutionalNeuralNetwork.Thetweetdatausedwere4403tweetswhichwillbeclassifiedinto5classes, namely: love, joy, anger, sadness, and fear. In addition,we conduct feature engineering todecide thebest feature inemotion classification.The featuresused in this researchareWord2Vec,FastTextandGlove.F1-scoreisemployedasanevaluationmetric.TheresultsofourexperimentsshowthatbyimplementingthecombinationofCNNandWord2Vecourdatasetcanachieve72.06%ofF1-Score,whichincreasesthebaselinemodelby63.71%
MappingComplexTouristDestinationPreferences:NetworkPerspectives
DianPuteriRamadhani,AndryAlamsyah,MuhammadNashirAtmaja,JoePanjaitanGiventherapidevolutionofinformationtechnologyincreasesthenumberofindividualtouristswhochoosetoenjoyfreetravelwithoutdependingonguidebooksandtouragencyservices.Informationsharingactivitiescontinuewithout limitation facilitatesmost tourists toexploreand independentlydecide their destinations based on online traveller review pages. This freedom generates anincreasingly complex tourist visiting pattern. In other side, the abundance of data provides a newapproachinanalyzingthisvisitingpattern.Wecollected215,168reviewswrittenbytouristsallovertheworldregardingtouristdestinationsinBali,Indonesia.Thisresearchanalyzestheonlinetravellerreviewdatathroughassociationruleminingtechniquestodetectpairsoftouristvisitingdestinationandmappedthemthroughthenetworkanalysisapproach.Weseparateourexplorationbytracingthetourist visiting movement as the underlying factor to understand tourist visiting behaviour. Wediscovered themostpopular touristdestinationvisitingpatternand thedifferences in tourist visitpreferences for each continent. This research contributes to support efficient mobility in tourismmanagementbyprovidingtourismdestinationnetworksinsight.
ADVANCED PROGRAM | ICOICT 2021
70
ABSTRACTSESSION2D:E-LearningandHCI
ExploringtheexistenceandvariationofGamePlayerTraitsamongUndergraduatestudentsin
Malaysia
MageswaranSanmugam
The use of game has become ordinary for students of generation-z whose live revolves aroundtechnology. The ease of access towards these games, via handheld devices such as tablets orsmartphones,cultivatesaspecifictypeofmotivationandaimsamongstudentswhenitcomestousinggames.Thesemotivationalgoalswill indirectlyinfluencewaysstudentsinvestigatetasksinvirtualformsthatmayormaynothaveasnippetofgame-likelearningenvironments.Assuch,itwillbevitalto identify thebreakdownofplayermotivationamongthesestudents.Participants froma tertiaryinstitutionweretestedusingBartlePlayerMotivationaftergoingthroughthelearningprocessviaane-learning. Itwas found that thehighest tendencywas forkiller typeplayersamong thestudents,followedbysocializers.Therewasadifferencewhen it came togenders; as such, the tendency toinvestigatetheneedsofindividualsplaysavitalrole.ImplementationofContinuousIntegrationandContinuousDelivery(CI/CD)onAutomatic
PerformanceTesting
DanaSulistyoKusumoandMohammadRizkyPratama
Performancetestisoneofthetestcomponentsinthedevelopmentphasetoensurethattheservicescreatedcanaccommodatethespecifiedtargetedperformanceloadandavoidperformancebottlenecks.Nowadays,withthedevelopmentofagiledevelopmentprocesses,thedevelopmentprocessescanrunfasteranditeratively.ContinuousIntegrationandContinuousDelivery(CI/CD)aremethodsusedinAgiledevelopmenttoautomateandspeedupthefollowingprocess:building,tests,andvalidationofservices.ThisresearchaimstoimplementCI/CDintheperformancetest.Intheexistingtest,theteststillneedshumanstoconductthetest.OurproposedsolutionforperformancetestswithCI/CDcanbeperformed automatically and reduce human role in the test. The implementation of CI/CD in theperformancetestmakestestprocessesintegrated,automatically,andperiodicallyexecuted.Itcanalsoquicklyrespondtoparametervaluechangesthatcanpreventthetesterfromsettingnewparametersinanewtestscenario.
UnderstandingGovernmentReorganizationImpactfromKnowledgeManagementPerspective:AStudyCase
YuliaSulistyaningsih,KhairiyahRizkiyah,SofianLusa,AssafArief
ReorganizationisacommonissuethatoftenoccursinIndonesia'sgovernmentinstitutionsthatcanincreasetheriskoflosingtacitknowledgeanddecreasetheagilityoftheinstitution'spublicservice.Properknowledgemanagement(KM)supportedbyadequateInformationTechnology(IT)canbeoneofthesolutions.However,thereissomeconsiderationrequiredtomakesuretheimplementationofKM can be done successfully. This study aims to take the initial step by evaluate the existingKMpracticeandassessKMimplementationreadinessusingaproposedmodelinoneofthegovernmentdepartmentsinXYZinstitution.Thisstudyusingquestionnairestocollectdatawhileinternalexpertsareinvolvedinvalidatingthequestionnaireandconfirmedtheassessmentresult.ThefindingshowsthatthereisstillnoKMstandardorpolicyappliesinthestudycase,andthereadinesshasreachedthelevelNotReadyNeedsSomeWorkwithascoreof2.909.Thesefindingscannotbeclaimedappliedto
ADVANCED PROGRAM | ICOICT 2021
71
allgovernmentinstitutions,butitcanbethepilotassessmentfortheinstitution,andthehighlightedissuecanbelistedascommonfactorstobewareofinsimilarinstitutions'conditions.
EnterpriseResourcePlanningTeachinginPostPandemicusingGamification
SitiFatimahAbdulRazak,FaizunizaMashhod,ZulfadhliNajmibinZaidan
TheglobalCOVID-19pandemichasseenariseindigitallearningmaterialsbeingsharedwithstudentsofalllevelsofeducation.Learninginstitutionsusuallyprovidealearningmanagementsystemwhereall the notes, tutorial and example past year examination questions are provided for students tosupporttheir learningactivities incoursesthroughouttheirstudies.Studentsusuallydownloadthelearningmaterialseitheras.ziporindividualfilesinvariousfileformats.Thestepsarerepetitiveforeachregisteredcourse, thereforecanbetimeconsumingforstudents.Studentsalsoneedtohaveasenseofappropriatefilemanagementskillinordertoorganisedownloadedmaterialsforeasyaccesswhenever necessary. When the number of courses grow throughout the years, improper filesorganisationmayresultinlostofaccessorunidentifiablefilesinstudentmachineordevices.Hence,thepurposeofthispaperistoinvestigatethepotentialofRoboticProcessAutomation(RPA)toaddressrelatedchallengesfacedbystudentsinmanagingtheamountoflearningmaterialsprovidedthroughalearningmanagementsystemorportal.ARPA-basedbotwasdevelopedandintegratedwithalearningmanagementsystemtoaccomplishthegoals.TheintegrationshowsthatRPA-basedbotscanminimisestudentseffortinmanagingtheirlearningmaterialsefficiently.
RPA-basedBotsforManagingOnlineLearningMaterials
KaushalJheengut,DineshaCauleechurn,BibiZarineCadersaibThe past decade has been witnessing a considerable technological evolution, gearing companiestowardsaconstantcompetitivesense.OneofthewaystomaintainacompetitiveedgehasbeentheadoptionofEnterpriseResourcePlanning(ERP)systemstostreamlinebusinessprocesses,thereforesettingahighdemandforERPprofessionals.AlthoughuniversitieshaveERPintheircurricula,thereisstillaskillmismatchregardingERPskillswithindustry.ExistingteachingmethodsforERPconsistsofnumerous challenges that must be addressed to ensure learning and skill acquisition take place.Furthermore, the impact of the COVID-19 pandemic accelerated the populace to revert to onlinebehaviors,includingonlineteaching.ThispaperproposesagamingapproachtoERPteaching,withtheaimofcomplementingexistingtraditionalteachingmethods,ifnotreplacethem.Thegameexposestheusertobasicconceptsthroughonlinelecturesandthenengagestheuserindifferenttypesofgamessuch as crosswords, quizzes, scenario-based games and a simulation game. This game wasexperimentedwithagroupofuniversitystudentsinMauritiuswhoalreadyhadexposuretoanERPrelatedsubject.Overall,thegamewasmuchappreciatedbythestudentsandtheywerepositiveaboutitsimplementationintheteachingcurriculumthatisgraduallymetamorphosingintodistancelearning.
ADVANCED PROGRAM | ICOICT 2021
72
ABSTRACTSESSION2E:Healthcare,Bioinformatics,and
BiomedicalApplication
Holick'sRuleImplementation:CalculationofProducedVitaminDfromSunlightBasedonUVIndex,SkinType,andAreaofSunlightExposureontheBody
JonathanSalomo,EduardusAriasena,AthayaSyaqra,SalmaMajidah
VitaminDisimportanttomaintaincardiovascular,skin,bone,andmentalhealth.Unfortunately,63%of Indonesianshave inadequate levelsofvitaminD.Oneof itsmajorcauses is the lackofpracticalmethodstomeasuretheamountofproducedvitaminDforeachindividual.Thisstudyconstructsapractical,mathematical formula tomeasure the amount of the produced vitaminD from sunlightexposure. It is conducted based onHolick's rule through skin type, body surface area exposed tosunlight, and ultraviolet (UV) index. Aided by correction fromUV spectrum analysis, the formulaenablespracticalmeasurementofvitaminDintakethroughsunlightforthegeneralpublic.Thefutureworksofthispapermayincludecorrectionsofmeasurementaccuracyandon-deviceimplementationofthealgorithm.
T-COFFEEMultipleSequenceAligneronHadoopSparkCluster
ViebiyantyPrihatiningrum,Setyorini,SitiAmatullahKarimah
DNA(DeoxyriboseNucleidAcid)isaseriesofnucleotideacidproteinsthatexistintheorganismbodywhere DNAwill be identicalwith inheritance. DNA in the organism body is in pairs, so biologicalanalysisisneededtomatchthesimilaritybetweentheDNAdata.SinceDNAcomputationusedhugeamountofdata,itshouldbecomputebyBigDatEnvironmenttocomputethesimilaritymatchingofdata.BigDataisusedforlarge-scalecomputationbyhavingseveralframeworksthatsupportsearchingforbiologysequencesimilarities.HadoopisaframeworkwhichisveryappropriateforrunningBigData.Inthisstudy,weusedMSA(MultipleSequenceAlignment)whereoneofthealgorithmswhichhas a high accuracy value is T-COFFEE (Tree Based Consistency Objective Function for AlignmentEvaluation) algorithm.T-COFFEE is an algorithm formultiple sequenceswhich is very suitable forfinding similarities in DNA data by focusing on very high accuracy values. Besides having a highaccuracyvalue,T-COFFEErequiresaverylongtimetoprocess.SothisresearchdidimplementationofT-COFFEEonhadoopparallelizationusingSparkwhichhasbeenproventoreducetheexecutiontime.
ADVANCED PROGRAM | ICOICT 2021
73
RelaxationOscillatorUsingClosed-loopDualComparatorforBiomedicalApplications
TheodoraValerie,DodiGarinto,PraptoNugroho,ArySyahriar,HarkiYanto
People'srisingconcernsabouthealthhasfueledresearchintohigh-performancebiomedicaldevicesandcircuits,suchassensorsandimplantablesystemsonachip.Itisimportantforthesebiomedicaldevices tooperateat thesame lowfrequencyto treatcertainorgansanddiseases.Thus, theclockgenerator needs to be able to provide low frequency clock pulses with minimal size and highreliability.Theseconstraintselevatepowerconsumptionandproductioncostwithinthecircuitdesignto key parameters. One promising candidate is relaxation oscillator which have good on-chipcompatibility,andsuperiorfrequencystability.Thispaperproposesanovelrelaxationoscillatorusingclosed-loopdualcomparator.Afrequencyinthebandof50Hzto2.5kHzcanbegeneratedbyreplacingthevalueof resistorsandcapacitor inallpossiblecombinations.Thesimulationandexperimentalresultconfirmthatclosed-loopdualcomparator-basedrelaxationoscillatorprovidelowfrequencieswithlowercostandmoresimplicityduetolesscomponents.
WirelessProgrammablebodysensornetworksandSituatedHealthcare
AlbertoFaro,DanielaGiordano,MarioVenticinqueWirelessBodySensorNetworksrepresentan interestingchallengetoeffectivelycontroldangerouspathologyandtomonitorthewellbeingstatusofelderandfragilepeople.However,suchnetworksdon'ttakeintoaccountthecontextthathighlyinfluencethehealthstatus.Thiscounteractswiththenotionofsituatedhealthcaremoreandmoreclaimedfrombothsociologicalandmedicalpointofviewto provide effective healthcare interventions. Following this point of view, the paper illustrates aneffectivearchitecturetocontrolhealthstatusinthecontextofthebodilyactivityandtheenvironment.Inparticular,programmablewirelessbodysensornetworksrecentlyavailableonthemarket fusedwithenvironmentalportablesensorsareevaluatedtoprovideaviablesteptowardsasmartsituatedhealthcare,i.e.,monitoringandcontrolfromremotesitesandinrealtimepeopleintheircontextbyusingIoTdevicesprovidedwithartificialintelligence,i.e.,AIoTdevices,andsuitablewebservicesthatallowsuchsmartbodilynetworkstointernetworkwithnetworksofotherpeople,doctorsorhospitals.Currentlimitsandpotentialitiesarepointedoutbyacasestudy.
ADVANCED PROGRAM | ICOICT 2021
74
ABSTRACTSESSION3A:Networking,IoT,andSecurity
DetectionofSinusoidswithFrequencyDriftinWhiteGaussianNoise
BradleyComar
In this study, 4 methods of detecting sinusoid targets in white Gaussian noise are investigated.AveragingFFTscoherentlyandnon-coherentlyisexamined.Abruteforcemethodisintroducedandinvestigated.UsingasinglelargeFFTspanningtheobservationwindowisalsoconsidered.Detectionperformanceisstudiedontargetswhosefrequenciesdriftovertime.Itisdeterminedthataveragingnon-coherentFFTsisasafechoiceinadetectionsystem.ThepreferredsizeoftheFFTusedfortonedetection aswell as the number of spectra to average is driven by the nature of the target tonefrequencydrift.This informationisespeciallyusefulwhenhighsampleratedetectorsareusedtofindtonesfromlowqualitytransmittersbecausemakingFFTstoolargeortakingtoomanyaveragesmayhinderdetection.
ToneDetectionSystemDesignforTargetswithFrequencyDrift
BradleyComarInthisstudy,3methodsoftonedetectionareinvestigated,thenon-coherentaveragingmethod,thepowerspectralmethod,andthecross-powerspectralmethod.Thetargettonesaresimulatedwithfrequencydrift.AnalysisisperformedtofindoptimalFFTsizesandnumberofspectratoaverage.PDvs.PFAcurvesare thencreatedandusedtocompare thesemethods.Thisanalysis isparticularlyusefulwhenusinghighsamplingratedetectorstofindtonesproducedwithlowerqualityoscillatorsthatexperiencefrequencydriftsinceusingFFTsthataretoolargeoraveragingtoomanyspectramayactuallydecreasedetectability.
AADC3:Active-ActiveDistributedControllerwith3-in-1AsynchronousHeartbeatSynchronizationMethodinSoftware-DefinedNetworks
MuhammadAriefNugrohoandVeraSuryani
TheproblemofDistributedControllers(active-active)usingsynchronousmessageexchangeisthateverymessagesentbycontrollerA(sender)willberespondedtobycontrollerB(receiver),thereforetheprocessofsendingthenextmessageafterreceivingthatresponse. It isCausingadecrease inperformance and an increase in the controller's workload because every message that will berespondedtorequiresanimmediateprocesstoproduceacknowledgments.Therefore,wedevelopedamethodtoimprovethemessageexchangemechanismandreducemessageresources'burdeninsending synchronization message information between controllers. This study proposes usingasynchronousmessagesasdistributedSDNcontrollermessageexchangeswiththe3-in-1method,which is the mechanism carried out by sending three messages and producing a reply in oneacknowledgment.Ourexperimentsshowthat3-in-1methodgivesloverCPUandmemoryusageandhasahigherthroughputthanthetraditionalmessageexchangeinDistributedSDNController.
ADVANCED PROGRAM | ICOICT 2021
75
4GLTECellularNetworkCoveragePlanningandSimulationonMandalayAreawithPropagationModelCost-Hatt
AhmadIdrisandSuciRahmatia
Cellularnetworkisexpensive,beitthetechnologyorthesupportinginfrastructure.Inordertoachievemaximumresourceefficiencyandeffectiveness,aplanisneeded.Toestimatemostappropriatecellthroughput,todecreasethenumberofequipments,toanswertheneedfortrafficanalysis,andtogetthemost optimal capacity network planning and simulation are required. Network planning is aprocess that consists of several activitieswhose final target is todefine anoptimal cost- effectivenetworkdesign,whichisthenbuiltasamobilecellularnetwork.ThisresearchwasmadeinMandalayarea,locatedinMyanmarwiththetotalsizeofcomputationalzoneis656,198km²whileaimingtohaveabetterunderstandingonplanninganddoingsimulationonsaidregion.WithCost-Hattamodelpropagation, CVVPX310R1 antenna type, and E-UTRA Band 3 - 20MHz frequency band, 58 LTEtransmitterareplaced..
USBFlashDrivesForensicAnalysistoDetectCrownJewelDataBreachinPT.XYZ(CoffeeShop
Retail-CaseStudy)
DanielSeptiantoUSBflashdrivesisusedwidelyasportablestoragedevicesandbecomepopularchoicetostoredortransfer the data among the employee. There was a greater concern about leaks of informationespeciallycompanycrownjewelorintellectualpropertythroughUSBflashdrivesduetotheft,lost,negligence, or fraud. This research is a case study within PT. XYZ company which aims to findremaininginformationrelatedtothecompanycrownjewelorintellectualpropertyinsidetheUSBflashdriveswithincompanyenvironment.Theresearchshowedthatthesensitiveinformation(suchascompanycrownjewelorintellectualproperty,andcustomercreditcarddata)couldberecoveredfromtheUSBflashdrives.Itcouldobtainhighriskimpact(reputational)tothecompanythathaslowsecurityawareness.
ADVANCED PROGRAM | ICOICT 2021
76
ABSTRACTSESSION3B:ComputerVision
EnhancedAlexNetwithSuper-ResolutionforLow-ResolutionFaceRecognition
JinChyuanTan,ChinPooLee,KianMingLim
Withtheadvancement indeeplearning,high-resolutionfacerecognitionhasachievedoutstandingperformancethatmakesitwidelyadoptedinmanyreal-worldapplications.Facerecognitionplaysavitalroleinvisualsurveillancesystems.However,theimagescapturedbythesecuritycamerasareatlowresolutioncausingtheperformanceofthelow-resolutionfacerecognitionrelativelyinferior.Inviewofthis,weproposeanenhancedAlexNetwithSuper-ResolutionandDataAugmentation(SRDA-AlexNet)forlow-resolutionfacerecognition.Firstly,imagesuper-resolutionimprovesthequalityofthelow-resolutionimagestohigh-resolutionimages.Subsequently,dataaugmentationisappliedtogeneratevariationsoftheimagesforlargerdatasize.AnenhancedAlexNetwithbatchnormalizationanddropoutregularizationisthenusedforfeatureextraction.Thebatchnormalizationaimstoreducetheinternalcovariateshiftbynormalizingtheinputdistributionsofthemini-batches.Apartfromthat,thedropoutregularizationimprovesthegeneralizationcapabilityandalleviatestheoverfittingofthemodel. The extracted features are then classified using k-Nearest Neighbors method for low-resolutionfacerecognition.EmpiricalresultsdemonstratethattheproposedSRDA-AlexNetoutshinesthemethodsincomparison.
AnEnd-to-EndOpticalCharacterRecognitionPipelineforIndonesianIdentityCard
AndreasChandraandRubenStefanusOpticalCharacterRecognitionhasbeenlongstudiedoverthepastfewyears.Thechallengeremainsforthespecificpurposeofextractinginformationfromimagedocuments.Theaimofthisstudyistocreateanend-to-endpipelineforanIndonesianidentitycard.AdeeplearningapproachwasusedtolocalizeanareaoftextinterestbyusingFasterR-CNNwithResNet-50asthebackbone,YOLOv5forcharacterdetection,andcombinedmachinelearningalgorithmsusingtheRandomForestalgorithmtoclassifycharacters.Theproposedpipelineshowedaremarkableresultforbothidentitynumberandfullname.Thisprovidesapowerfultoolfortheauto-fillformandverificationprocesseffectivelyandefficiently.
AStudyofBatikStyleTransferusingNeuralNetwork
AdityaFirmanIhsanInthisstudy,tworemarkableapplicationsofconvolutionalneuralnetwork,i.e.texturesynthesisandstyletransferareappliedtobatiktexture.Individuallayersfromfourpre-trainednetworkssuchasVGG-19, Inception V3, ResNet-50, and DenseNet-121 are compared and analyzed. Different batikmotifswithsomespecificcriteriaarealsocomparedtoseethecapabilityoforiginalstyle transferalgorithm to regenerate concrete texture of Batik. Lastly, we propose a way to reconstruct batikimageswithsomepatternsfollowinganobject'sshapecontainedincontentimage.
TomatoPlantDiseaseIdentificationthroughLeafImageusingConvolutionalNeuralNetwork
AuliaIkvandaYorenandSuyantoSuyanto
Theproblemthatoftenoccursinagricultureisaboutdiseasesinplants.Plantdiseasescanresultinreducedyieldsfromagriculturalproduction.Therefore,thedetectionandanalysisofplantdiseasesare
ADVANCED PROGRAM | ICOICT 2021
77
criticalandshouldbedoneasearlyaspossible.Diseasesinplantsoftenappearontheleaves,andthecharacteristicsof theaffected leavesareverydiverseanddifficult todistinguish.Thisphenomenonresultsindifficultyintheidentificationofplantdiseasesautomatically.Oneofthetechnologiesthatcanbeusedinidentifyingleafproblemsisdigitalimageprocessingtechnology.Theplantusedasacasestudyinthisresearchisthetomatoplant.AlternariaSolani,Septorialeafspot,Yellowvirusaresomeofthedisordersthattomatoplantscanexperience.Thesedisordersshouldbeclassifiedaccordingtotheirtype.Thisresearchdesignsasystemtoclassifythreetypesofdiseaseexperiencedbythetomatoplantleaves.Adatasetof4400 leaf images iscollectedand learnedto theConvolutionalNeuralNetwork(CNN)toclassifythreetomatoplantproblemsusingtheAugmentationprocess.Anevaluationusing5-foldcross-validationshowsthatCNNwithaugmentationdatagivesanaverageaccuracyof97.8%andthehighestaccuracyof99.5%.Thisresultisbetterthanthepreviousmethods:AlexNet,FasterR-CNN,andCNN+redgreenblue(RGB).
ADVANCED PROGRAM | ICOICT 2021
78
ABSTRACTSESSION3C:DataScience
Non-StationaryOrderofVectorAutoregressioninSignificantOceanWaveForecastin
FikkaRaudiyaandAniqAtiqi
Thispaperstudiestheimplementationofnonstationarymultivariatetimeseriesmodeltofitthewaveoceandata.Amodelwhichcomprisesaregressiontermandanassociationwithexogenousvariablesinaparticulartimehorizon.Becauseofthetrendfluctuationinthedataleadingtounstableprocess,differenceddataareusedinfittingthemodel.TheapproachsuggestedisappliedtothefiniteorderofVector Autoregression for an improved prediction simultaneously ofwave ocean by carrying outwind-related information to waves. The proposed model is compared with linear simpleAutoregressivemodel.Theperformanceofbothforecastingproceduresisassessedbymeansofwell-knownerrormeasures.The forecast basedon theproposedmethodology indicated that it canberegarded as a promisingmethod forwave ocean prediction, it outperforms using 4-order VectorAutoregression.
FN-Net:ADeepConvolutionalNeuralNetworkforFakeNewsDetection
KianLongTan,ChinPooLee,KianMingLim
Information and communication technology has evolved rapidly over the past decades, with asubstantialdevelopmentbeingtheemergenceofsocialmedia.It isthenewnormthatpeoplesharetheirinformationinstantlyandmassivelythroughthesocialmediaplatforms.Thedownsideofthisisthat the fake news also spread more rapidly and diffuse deeper than before. This has caused adevastating impact on people who aremisled by the fake news. In the interest of mitigating thisproblem,fakenewsdetectioniscrucialtohelppeopledifferentiatetheauthenticityofthenews.Inthisresearch,anenhancedconvolutionalneuralnetwork(CNN)model,referredtoasFakeNewsNet(FN-Net) isdevised for fakenewsdetection.TheFN-Netconsistsofmorepairsofconvolutionandmaxpooling layer to better encode the high-level features at different granularities. Besides that, tworegularization techniquesare incorporated into theFN-Net toaddress theoverfittingproblem.ThegradientdescentprocessofFN-NetisalsoacceleratedbytheAdamoptimizer.TheempiricalstudiesonfourdatasetsdemonstratesthatFN-NetoutshinestheoriginalCNNmodel.SentimentAnalysisofOjekOnlineUserSatisfactionBasedontheNaïveBayesandNetBrand
ReputationMethod
AlamRahmatulloh,RahmiShofa,IrfanDarmawan,ArdiansahArdiansah
GojekandGrabarethemostpopularonlinemotorcycletaxisandareoftenusedtodayinIndonesia,basedonHootsuite'ssurvey.However,itisnotyetknownhowtheresponsefromonlinemotorcycletaxiusers.SoitisnecessarytohaveasentimentanalysisofonlinemotorcycletaxiuserswhethertheyaresatisfiedordissatisfiedwiththedriversandGojekandGrabcompanies'services.Twitterwith52%activeusersofallinternetusersinIndonesiaallowsuserstowritevarioustopicssothattofindoutthelevelofusersatisfactionwithGojekandGrab.SentimentanalysiscanbeusedasareferenceforthedevelopmentofGojekandGrabservicesinthefuture.TheymeasurethelevelofsatisfactionwiththeNetBrandReputation(NBR)methodfromtheNaïveBayesclassificationresultsusingtherapidminertool.Theratingwithaccuracyhasanaccuracyvalueof99.80%forGojekand99.90%forGrab.ThisstudyshowsthatmoretweetshavenegativeopinionscomparedtopositiveopinionsforGojekandGrab.Namely616positiveopinionsand2317negativeopinionsforGojekdrivers,3560positiveopinionsand6419negativeopinionsforGojekCompany.594positiveopinions,and1866
ADVANCED PROGRAM | ICOICT 2021
79
negativeopinionsforGrabdrivers.Aswellas3516positiveopinionsand4407negativeopinionsforGrabComp.
RawPaperMaterialStockForecastingwithLongShort-TermMemory
FebryoKurniawan,DyahHerwindiati,ManatapDolokLauro
The manufacturing business is one of the businesses in Indonesia that continues to show itsdevelopmentfromyeartoyear.Likeamanufacturingbusinessingeneral,oneoftheimportanteffortsmade in theprintingbusiness is the supplyof rawpapermaterials toproduce finishedgoods.Thepurposeofthisresearchismakingaforecastingoftherawpapermaterialforprintingcompanyon7differenttypesof269historicaldatawithweeklyintervalsfromJanuary2015toFebruary2020beforetheCovid19pandemicseason.ForecastingisdoneusingtheLongShortTermMemorymethodwithPythonlanguage.ThemodelarchitecturefortrainingandtestingiscarriedoutusingvanillaLSTMwithsingleinput,hiddenandoutputlayerwiththeconfigurationof64neuronsinthehiddenlayer,150epoch,12batchsizeandAdamOptimizer(lr=0.0001)whichwasrepeated10timesforbestresult.Thetestresultsshowthebestwindowsizelengthinthemodelforeachpaperrawmaterialdifferentlyfrom4to16.AllmodelswassuccessfullyforecastingthetestdatawithanaverageMAPEoftheoverallforecastof21.48%.
MobileCustomerBehaviourPredictiveAnalysisforTargetingNetflixPotentialCustomer
SuryadiTanuwijaya,AndryAlamsyah,MayaAriyantiThe development of Indonesia's ICT environment has made the mobile video-on-demand (VOD)platform one of the emerging lifestyles. With advanced smartphone technology, mobile phonesubscribers able to enjoy high-resolutionmobile VOD servicewith a greater user experience. ThepurposeofthisstudyistoprofileandpredictpotentialcustomersofoneoftheVODplatforms,Netflix,for personalizing marketing targets. Using machine learning predictive analytic methodology,customerprofileandbehaviordataaredividedinto3clustersusingtheK-Meansmodelbeforetestedwithseveralsupervisedmodelsforgettingthebestmodelforeachcluster.Featureimportanceanalysisisconductedtosupportmarketinginsightforproductofferingfollowsuptoeachtargetedpotentialcustomer.SignificantvariablesaffectingNetflixbuyersandnon-buyerswithin3differentclustersaredefined clearly with the number of potential customers that can be targeted as Netflix's futuresubscribers. The result shows themethod can be used by themobile operator to target potentialcustomerswitheffectivepromotionalorproductofferingbypersonalizedmarketingapproachbasedon the behavioral pattern and customer needs. It is expected by implementing this methodology,effectivityandaccuracyofmarketingeffortwillbeincreasedcomparedtotheconventionalmethod.
ADVANCED PROGRAM | ICOICT 2021
80
ABSTRACTSESSION3D:E-LearningandHCI
MasterDataManagementMaturityModel(MD3M)Assessment:ACaseStudyinSecretariatof
PresidentialAdvisoryCouncil
ChielsinKo,AndytiasAdywiratama,AchmadHidayanto
ImplementationofMasterDataManagement(MDM)inanorganizationaimstohelptheprocessofconsolidatingandintegratingvarioussourcesofmasterdataintoasolitarytruthsource.MDMisalsoable to help to solve the complexity of data that occurs in the synchronization, consolidation andcleaningprocessofdatafromredundancy,andtoincreasethevalueofoperationalefficiencythroughthestandardizedcontentwhichrepresentsbusiness-basedprocess.Thegoalofthisstudyistoassessthe level of MDM maturity at Secretariat of Presidential Advisory Council during the COVID-19pandemicera.TheassessmentwasdoneusingMasterDataManagementMaturityModel(MD3M)bySpruit-Pietzka.ThemasterdatamanagementmaturitylevelintheorganizationwasassessedusingaquestionnairefilledoutbySMEsinagroupdiscussion.TheresultoftheassessmentshowedthatMDMmaturitylevelatSecretariatofPresidentialAdvisoryCouncilis1.Thereare61.29%(38oftotal62)implementedcapabilities.Thismeansthatorganizationalreadyhasawarenessinthemanagementofmasterdata.TheorganizationcanimproveitsMDMmaturitytoahigherlevelbyimplementingthemissingcapabilities.CapturingInstitutionandLearnersReadinessofe-LearningImplementation:ACaseStudyofa
UniversityinBandung,Indonesia
DawamDwiJatmikoSuwawi,BayuAditya,NungkiSelviandro,AnisaHerdiani,YatiRohayati,YanuarFirdausArieWibowo
E-learningreadinessassessmentisacriticalprocessthataninstitutionneedstodoinimplementinge-learning.Byconductingane-learningreadinessassessment,aninstitutioncouldidentifythefactorsthathinderthesuccessfulimplementationofe-learninganddevelopastrategicplantoenhancethee-learning implementation continuously. The higher education institution (HEI) needs to adopt thisapproach to implement e-learning successfully. Previous studies, such as inTurkeyHEIs, EgyptianUniversity,andtheUniversityofMysore,India,reportedthate-learningreadinessassessmentprovidesbenefits in defining a future direction in e-learning innovation. Telkom University, one of the topprivateuniversities inBandung, Indonesia, has launchedmanypolicies andprograms related to e-learning to provide high-level academic services and increase productivity and efficiency in thelearningprocess.Someindividualshaveraisedsomeresistancetothesepoliciesandprograms,buttheactualdataregardingthisisminimal.Themanagementneedsdatarelatedtoe-learningreadinesstointervene andmake this e-learning program a success. This study aims to capture the e-learningreadiness inTelkomUniversity fromthe institutionaland learner'sperspective.Data for thisstudywerecollectedusinganonlinequestionnaireaswellasdepth-interviewasarchivalsources.Besidesidentifying the e-learning readiness index, the output from this study is also to providerecommendationstopolicymakersregardingtheimplementationandthedevelopmentofe-learninginTelkomUniversity.
SatisfactionFactorsofIndonesianNationalCivilServantRecruitmentSystem
GalihKenangAvianto,FitriaElliyana,DanaI.Sensuse
Citizensatisfactionisthekeytothesuccessofe-governmentservices.Thecomputer-basedselectionsystem is one of the Indonesian government's services for recruiting civil servant candidates.
ADVANCED PROGRAM | ICOICT 2021
81
Therefore,findingfactorsofIndonesianelectronicsatisfactionisgoaloftheresearch.Basedontheliteraturestudy,sixhypothesesandsixfactorswereidentifiedthatcandeterminethelevelofusersatisfactionwith the computer-based selection system. Surveyof1,070 respondent fromdifferentprovinces in Indonesiawasconducted.Dataanalysisusingmultipleregressionand factoranalysisshowsthatthefactorsthatinfluencecomputerassistedtestsystemofNationalCivilServiceAgency(BKN)usersatisfactionaresecurityandprivacy,accessibility,awarenessofservice,andtransparency.
ImplementationandAnalysisofReusabilityFrameworkDesignforEventUserInterface
ComponentinPhaser3
AhmadArsyelAbdulHakim,DanaSulistyoKusumo,JatiH.HusenThegrowthofgamedevelopmentisnowwidelysupportedbysoftwaresuchasgameenginesandgameframeworks.Phaser3isapopularHTML5basedgameframeworkonthebrowserplatform.However,Phaser3doesnotfacilitatemanagingcodeofUserInterface(UI)eventcomponent.Sometimesagamedeveloper discovers that UI event component codes, such as buttons or joysticks that have beencreated,must be regenerated on subsequent projectswith the same function. In this research,weproposedareusability framework forPhaser3bycombiningthe factorymethodandthesingletonpatterninmanagingUIeventcomponents.Theaimistomakedeveloperproductivityincreasedsothatthereisnoneedtorepeatthealgorithmorcodethathasbeencreatedpreviously.OurresultsshowthatthePhaser3UIstandardhaslessareusabilityvalueof64,241%,whiletheproposedframeworkthatwasbuilt has relatively higher reusability, that is equal to 84,576%.Gamedevelopers can add thecombination of design patterns, as good software development practices, into an existing gameframeworktoachievecodereusabilitywithoutmajorchangingontheexistinggameframework.Thisapproachcanalsobeusedforthecreatorsofgameframeworktousedesignpatternsfromtheoutsetofdevelopmentofgameframework.
ThePreliminaryStudyonthePerceptionofEngineeringStudentsonBlendedLearning
MinChiLow,ChenKangLee,ManjitSinghSidhu,ZaimahHasan,SengPohLim,SengCheeLimMechanics Dynamics is an important fundamental course in engineering education. However, thiscourseexperiencesahighfailurerateamongengineeringstudents.Thismaybeduetovisualizationproblemsassociatedwithstaticimages,complexengineeringmodels,andconceptmisunderstanding.ThispaperpresentsthepreliminaryresearchofstudentperceptionontheblendedlearningflippedclassroomapproachinovercomingtheirlearningdifficultiesinMechanicsDynamicscourse.Thispilotstudy aims to collect the learningdifficulties of students in theMechanicsDynamics and students'perceptionofblendedlearningusingflippedclassroomapproach.Aquestionnairehasbeendesignedand distributed through an online platform. The sample size is 30 which targets the engineeringstudentsinMalaysia'suniversitywhoalreadytooktheMechanicsDynamicscourseinlessthanfiveyears.Thefindingsareanalyzedusingadescriptivestatisticsapproach.Theinitialfindingsindicatethatthevisualizationproblemisthemainconcernamongthestudents.Althoughthestudentsshowlow awareness regarding the blended learning flipped classroom approach, they have a positiveattitudetowardstheelementoftheblendedlearningapproachtobeimplementedintheirclassroom.
ADVANCED PROGRAM | ICOICT 2021
82
ABSTRACTSESSION3E:Healthcare,Bioinformatics,and
BiomedicalApplication
ImplantSegmentationinRadiographicImageryUsingMultiresolutionMTANNandWaveletDecomposition
RanggaPerwiratama,PranowoPranowo,DjokoBudiyantoSetyohadi
Occasionally implants are coveredwith amusculoskeletal system inmedical imaging, it could bedifficultformedicalpersonnelsaswellascomputerstoperceivetheseimplants.Inthispaper,weuseimageprocessingtechniquestosuppressthemusculoskeletalsystemcontrastinX-rayimagesthatcontainimplantsthroughartificialneuralnetworkscalledmassivetrainingartificialneuralnetworks(MTANN).MTANNisanon-linearfilterthatcanbetrainedusingappropriateX-rayimagesinputandlearningimages.Learningimagesareobtainedthroughtheeditingprocessusingagraphicaleditor,whichisusedtocorrectlearningimagesfromnoise.Toeffectivelysuppressamusculoskeletalsystemthat has multiple spatial frequencies, we use multiresolution MTANN consisting of waveletdecomposition/compositiontechniquesanddedicatedMTANNforeveryresolution.MultiresolutionMTANNhopefullycanprovideimagesthatarereallyclosetolearningimages.ByreducingthecontrastofthemusculoskeletalsystemfromX-rayimages,wewillbeabletocreateX-rayimageswherethemusculoskeletalsystemissubstantiallysuppressedandbeableseetheimplantclearlyincontrast.WeuseMURAasatrainingdatabaseandforvalidationtests.MURAisalargesetofboneradiographicdata,which isdesignedto trainANNtodeterminewhetherX-ray imagesarenormalorabnormal.Once the model is done, when our technique is applied to non-training X-ray images, themusculoskeletal image system is substantially suppressed while maintaining the visibility of theimplant.OurimageprocessingtechniquetosuppressmusculoskeletalsystemsusingmultiresolutionMTANNwillpotentiallybeusefultomedicalpersonnels.ImprovingMulti-ClassMotorImageryEEGSignalsClassificationUsingEnsembleLearning
Method
DeniKurniantoNugroho,NoorAkhmadSetiawan,HanungAdiNugroho
Electroencephalography(EEG)isatechniqueformeasuringelectricalactivityonthescalp.TheEEGdetectsvoltage fluctuationscausedby ioncurrents inbrainneurons.Thebrain-computer interface(BCI)isintendedtoenablehumanstomonitormachinesandinteractwithcomputersthroughtheirbrains. It intends to construct non-muscular production pathways to convert brain function intodiscriminatorycontrolcommandscorrelatedwithvariousEEGsignalsdependentonmotorizedimagepatterns.ResearchonEEGiscurrentlygrowing,especially inthefieldofmotorimaging.EEGsignalprocessingwouldbeafeasibleoptionfordevelopingsuchaBCIdevice.ThefourbasicstagesinclassicalBCI are multi-channel EEG signal acquisition, signal preprocessing, feature extraction, andclassificationofmotorimagepatternsbasedonEEG.Thisstudyaimstodeterminetheeffectofwaveletpacketdecomposition(WPD)andcommonspatialpattern(CSP)featureextractiontooptimizefeatureselectionusingtheensemblelearningmethod.Themethodusedinthisresearchisexperimental,wherethe stages begin with preprocessing, feature extraction with WPD and CSP, classification usingensemblelearningandimplementingfeatureselectionusingtheprincipalcomponentanalysis(PCA)andselectfromthemodel(SFM).Theresultsarethecomparisonoftheaccuracygeneratedfromeachmethod, includingrandomforest(RF)of74.71%,randomforestwithprincipalcomponentanalysis(RFPCA)of68.01%,randomforestwithselectfromthemodel(RFSFM)of82.15%,extratrees(ET)of77.97%, extra trees with principal component analysis (ETPCA) of 64.18% and extra trees withselectedfromthemodel(ETSFM)of83.28%.
ADVANCED PROGRAM | ICOICT 2021
83
ImplementationandExperimentalCharacterizationofDual-BandWearableReflector
ComposedofAMCStructureforWirelessCommunication
AchmadMunir,DwikiHaryanto,IchsanNusobri,LevyOliviaNur
This paper presents an implementation of dual-band wearable reflector composed of artificialmagnetic conductor (AMC) structureand its characterization throughexperimentalmeasurement.Theproposedwearablereflectordesignedforwirelesscommunicationsisconfiguredby3x3unitcellofAMCstructure.Eachunitcellisconstructedbyacombinationofasquarepatchandasquareringpatch with gaps arranged concentrically. An RT/Duroid RO3003 dielectric substrate with thethicknessof0.5mmisemployedforthedesignaswellastheimplementation.Meanwhile,aprinteddipoleantennaasaninseparablepartofwearablereflectorisalsodeployedonthesamedielectricsubstrate.Thisconfiguration,i.e.,aprinteddipoleantennaandawearablereflector,isexpectabletooperateattwodifferentfrequencybands,ordual-band,coveringtherequiredfrequenciesofwirelesscommunications while maintaining its compact size and simple configuration. Experimentalcharacterization is applied to validate the proposed design, inwhich the results demonstrate theabilityofconfigurationtoworkattwofrequencyresponses,namely2.45GHzbandand3.35GHzband.
ADVANCED PROGRAM | ICOICT 2021
84
ABSTRACTSESSION4A:Networking,IoT,andSecurity
ModifiedPixelValueOrdering-basedPredictorforReversibleDataHidingonVideo
TohariAhmad,AlekNurFatman,AhmadHoirulBasori
Communicationsecurityhasbecomeanissuewheremassivedataexchangestakeplace.Hidingsecretdataincertainfileduringtransmissioncanbeapotentialwaytoprotectthosedata.Nevertheless,someparameters,suchtheresultedstegofilequalityandthedatacapacity,arestillchallenging.Inthispaper,amethodthatmodifiedPixel-basedPixelValueOrdering(PPVO)Predictorisproposedforreversibledatahidingonvideo.PPVOcanembedthepayloadintoapixelwhosevalueisequaltothemaximumorminimumcontextpixel.Toincreasethenumberofpayloadsthatcanbeembedded,wemodifythePPVOmethodbyusingonlybothrightandbottompixelstodevelopthepredictors.Embeddingprocessiscarriedoutiftheconsideredpixelhasthesamevalueasthatintherightorbottom.Theexperimentalresult shows that this method has a significant increase in the embedding capacity and quality.Nevertheless, the overall maximum capacity of each cover is less than some others, while itscorrespondingqualityisstillthehighest.SimulationandAnalysisofPartialTransmitSequenceonPalmDateLeafClippingforPAPR
ValueReduction
Vincent,AntoniusSuhartomo,JoniW.Simatupang,MiaGalinaTheusageofFFTthathasbecomethekeyconceptinOFDMsystemproducesahighPAPRvalue.Inorderreduceit,thereareseveraltechniquesthatcanbeimplemented,suchasthePalmDateLeafclipping,andthePartialTransmitSequence.Previousresearchershaveevaluatedeachtechniqueindividually.Inthisresearch,theAuthorsevaluatethePAPRvalueastheeffectofaddingPartialTransmitSequencetothePalmDateLeafclippingtechnique.Theevaluationisdonewithseveralmodulationtechnique,suchasQPSK,8-PSK,16-PSK,8-QAM,and16-QAM.SincelowPAPRperformanceisnotadvantageousifthesignal'sBERvalueisworsen,thustheevaluationalsoconsiderstheBERperformanceofthesignal.Inthiscase,theAuthorfocusesontheBERperformanceoverAWGNchannel.Theresultshowsthatinallofthescenarios,PTStechniquecouldimprovethesignal'sBERandPAPRperformancefora lowCRvaluesuchas5dBand7dB.Additionally,forthehigherCRvaluesuchas10dBand20dB,thesignal'sBER performance is similar with the normal OFDM signal. Even so, it provides a consistent PAPRreduction of approximately 3 dB. This way, PTS technique always provide improvement in BERperformanceof thesignal.As for thePAPRperformance,PTS technique isable to improveallcasesexceptfor8-QAMand16-QAMsignalwithclippingtechniqueatlowclippingratiosuchas5dB.
DesignAutomationofSinglePhotonCountingMethodforQuantumRandomNumberGeneration
DwiNovazrianto,AlwanMuhamadFajar,MuhammadYusuf,ApriliaKusumaDewi,RiniWisnu
Wardhani,DedySeptonoCaturPutrantoEDU-QCRY1isoneofthequantumkeydistributiondevicesinordertogeneratequantumrandombit.Inthispaper,weconductamechanismtoproduceakeysequencebasedonsinglephotoncountingusingEDU-QCRY1SinglePhotonDetectormoduleforreceivingdatawhichcanbeacquiredusingthereceiver's device automationusing a LightDependentResistor parameter. The design of quantumrandomgeneratorisusingprotocolofQuantumKeyDistribution(QKD)thencomparedtotruetableoftheBB84protocol.Theaveragespeedofthequantumbitintheautomationdeviceis0.0604secondswithestimationfor106quantumrandombitaround16hoursexperiment.Withintherandombits,
ADVANCED PROGRAM | ICOICT 2021
85
somedifferenceoccurswhenthesenderandreceiverhaveadifferentbasis.Thisresearchprovestheefficiency of automation bit storage on the EDU-QCRY1 Single Photon Detector for generating aquantum random bit and has the valid result as the experiment in performing key distributionsimulationsusingEDU-QCRY1devices.
OntheModificationsofaDigitalSignatureAlgorithmwithSecretSharing
MaretaWahyuArdyaniandUmiUliZulfahCryptographicalgorithmsarevulnerablenotonlyfromattacks,butalsofromtheissuesraisingfromstoringtheprivatekey.Oncetheprivatekeyisleaked,nomatterhowstrongthealgorithmsareclaimedtobe,weareseverelycompromisedandhence,losethesecrecy.Oneofthesolutionstoremedythissituationistoemploythesecretsharingscheme.Thebasicideaofsecretsharingistosplitoneprivatekeyvalueintoseveralsharevalues.Thosesharevaluesarethenstoredintodifferentplacesbydifferentsecretshareholders.Thepresenceofonlyonesharevaluecannotrepresentthewholesecretvalue,andfurther,isunabletoreconstructtheprivatekey.Inthisstudy,weinsertedasecretsharingschemeintoadigitalsignaturealgorithmbasedonconiccurvecryptography.Weoptedtoperformtheschemeonconicbasedalgorithmbecausethesecurvesarebelievedtohavesimplercomputationsthanthoseofelliptic curves.We split the private key value into several share values,which are then utilized togeneratesignaturevalues.Thealterationonsignaturevaluegenerationprocess (byaddingasecretsharingscheme)doesnotaltertheverificationprocesstakingplaceontheoriginalalgorithm(withoutthe secret sharing scheme). We also employed additional mechanism by using the solution ofcongruencesystembasedontheChineseRemainderTheorem(CRT)andFermat'sLittleTheorem.Thewholeprocessisexpectedtoprovideifnotlayersofsecurity,amechanisminwhichwecanmitigatethepossibilityoflosingsecrecy.
ConnectedVehicleCommunicationConceptforFloodLevelWarningUsingLowCostMicrocontroller
MohdFikriAzliAbdullah, SumendraYogarayan,SitiFatimahAbdulRazak,FremontKwong
InMalaysia,therearesomeareasthatfrequentlyhavefloodduetoditchblockageorsuddenrisingofriver'swater.Mostfloodscancausevehiclesandotherinfrastructurestobedamaged.Thevictimsofeveryfloodthatoccurcouldbereducediftheyreceiveanymessagethroughacommunicationmediumtoalertinadvance.Therefore,thisstudyaimstodevelopaprototypesystemtosendamessageonflooddetection via wireless connection using ESP8266 as a means of communication. This prototype isspecifically applied for vehicles to bewarnedmuch earlier. The outcome shows that sending flooddetectionmessagebyusing the communicationmedium is possible and took less than aminute totransmitthemessage.TheflooddetectiondataaresavedinFirebasecloud.
Randomness,Uniqueness,andSteadinessEvaluationofPhysicalUnclonableFunctions
ParmanSukarno,RivaldoLudovicusSembiring,RizkaRezaPahleviThedevelopmentoftheInternetofThings(IoT)canbefoundinvariousplaces.However,multiplekindsof attacks have also increased. IoT devices are very vulnerable to attacks, both physical and non-physical,becauseof theirunmannednature. Innon-physicalattacks, themost important thing is tosecurethedataonmemorydevices.Physicalunclonablefunction(PUF)isthestrongestandlightestmethodtosecuringmemorydevicesandcanbeusedonunmannedIoTdevices.TheadvantageofPUFovercurrentclassicalcryptographytypes is itscompatibilityonIoTdeviceswith limitedcomputingresources. However, before PUF can be claimed to provide security property, it must meet theevaluation indicators: randomness, uniqueness, and steadiness. PUF can be the best solution for
ADVANCED PROGRAM | ICOICT 2021
86
securingdataonIoTdevicesbecausetheencryptionprocessdoesnotputasecretkeyonthedevice.Instead,thekeyisgeneratedrandomly.ThisresearchisevaluatingtwodifferentPUFchipswiththesamePUFdesign.WedesignedthearbiterPUFontheFPGAandevaluatedtheresultsoftheresponsesgiven.Throughrigorousexperiments,thisresearchsucceededtoevaluatethethreeindicatorsofPUFwheretherandomnessis54.43%:45.4%,and25.88%:74.2%,theuniquenessbetweenchipis69.53%,andlastly,thesteadinessis89.84%,and91.41%.
ADVANCED PROGRAM | ICOICT 2021
87
ABSTRACTSESSION4B:ComputerVision
ALow-CostHigh-AccuracyThermalCameraUsingOff-the-shelfHardwareDevices
Dinh-TienTran,Viet-KhoaNguyen,Ngoc-ThienNhan,Duc-HiepNguyen,Hoang-AnhPham
TheCovid-19pandemicisspreadingworldwide,thedemandforcontrollingandscreeningfeverisincreasing very rapidly. Some traditional methods like using thermometers or medical checkingcannot handlemany people in crowded areas such as airports, train stations, or buildings. Somethermal imaging systems are used to control people's ins and outs. These systems help manygovernmentsdecreaseandpreventviruses'spreadbyscreeninghumantemperaturebasedonskinradiation.Despitealloftheadvantagesthistechnologyhashad,thissystem'sveryhighcostisonereasontopreventsomeorganizationsfromapplyingitdespitethehighpotentialofimageprocessing,artificialintelligence,andtheInternetofThings.Thispaperwillpresentasystemcombiningalow-cost thermal camera and a standard RGB camera for screening fever combined with attendancechecking.Theexperimentalresultsshowthatthisproposedsystemcanbeappliedinmanyareas,fromuniversitiestobuildings,withtheminimumcostandacceptableperformance.
Sentinel1ClassificationforGarlicLandIdentificationusingSupportVectorMachine
MuhammadAsyharAgmalaro,ImasSukaesihSitanggang,MiaWaskitoThehighdemandforgarlicisnotcomparablewiththeresultsofdomesticgarlicproduction.Indonesiangarlicneedsfulfilledbyimportsupto95%ofnationalneeds.TheMinistryofAgriculturehasaprogramofthecultivationofgarlicinSembalun,EastLombok,WestNusaTenggarainordertorealizegarlicself-sufficiency.ThisstudyaimstoidentifythegarliclandinSembalunusingtheSentinel1Asatelliteimage.Theimageconsistsofdual-polarizationVVandVHvalues.ImageswereacquiredinJulyandNovember2019fortheareaofSembalun,EastLombok,WestNusaTenggaraIndonesia.Preprocessingdatastepsinvolve applying orbits, calibrations, speckle filters, terrain corrections, and linear to dB. SupportvectormachinealgorithmisusedtoclassifySentinel1Aimages.Hyperparametertuningwasdonetogetthebestparameterswhichareregularizationparameter(C)10,gamma1,andtheRBFkernel.Theclassificationmodelhasaccuracyof76%,precisionof71%andrecallof89%.
RecognitionofAcademicEmotionsinOnlineClasses
JordanMingHanPang,TeeConnie,GohKahOngMichaelOnlineeducationhasproliferatedsincetheCOVID-19pandemic.Classeshavebeenmovedonlineasaresultofschoolclosures.Despitetheflexibilityofferedbyonlinelearning,thereareseveralchallengesfaced.Creatingagoodclassroomclimateforonlineclassesisachallengingtask.Itisdifficultfortheteacherstoobtainemotionalfeedbackfromthestudents,especiallyinasynchronousclassesorclasseswithlargenumberofstudents.Itishardfortheteacherstoevaluatetheengagementofthestudentsinclasswithoutknowingthestudents'emotionalresponse.Theexistingfacialexpressionrecognitiondatabasesfocusonbasichumanemotionslikehappy,angry,sad,surpriseandneutral.Thesebasicemotionsarenotappropriateforlearningaspsychologicalandpedagogicalstudieshaveshownthatthere are differences between basic human emotions and academic emotions. In view of theseproblems, thispaperpresentsa studyonacademicemotions.Adataset comprising fourpertinentacademicemotionshavebeenestablished.Empiricalanalysisonthedatasetisconductedusingbothhandcraftedanddeeplearningapproaches.Thebaselineevaluationdemonstratesthesuitabilityoftheestablishedacademicdatasetforonlinelearning.
ADVANCED PROGRAM | ICOICT 2021
88
ImageSteganographyCompressiveSensingOrthogonalMatchingPursuit
IrmaSafitri
Steganographyistheartofwritingorhidingmessagesinacertainwayonadigitalfile.Thepurposeofsteganographyisthatamessagecanonlybeseenbythetworelatedpartieswithoutbeingnoticedbytheotherperson. In thisstudy,asteganographysystemwasdesignedusing imagemedia tohideasecretmessage.Imagecompressionisfirstperformedusingcompressivesensing(CS)sothatanimagewith a smaller resolution is obtained. Stationary wavelet transforms (SWT) are used as thetransformationmethodandsingularvaluedecomposition(SVD)isusedastheinsertionmethod.Toreconstructorreturnthecompressedmessagetotheoriginalmessage,orthogonalmatchingpursuit(OMP)algorithmisused.Thesystemperformancewastestedusingsalt&peppernoiseattackandgaussiannoise.ThetestresultsshowthattheBERvalueis0.05765andthePSNRvalueis105.666dBin thesteganographysystemwhenattackedwithsalt&peppernoiseatadensitynoiseof0.01. Inaddition,oursystemalsoshowsaBERvalueof0.0509atsigma1andaPSNRof104.615dBwhenGaussiannoiseattacksareapplied.
ADVANCED PROGRAM | ICOICT 2021
89
ABSTRACTSESSION4C:DataScience
IndonesianHoaxIdentificationonTweetsUsingDoc2Vec
TitiWidaretna,JimmyTirtawangsa,AdeRomadhony
Inthispaper,wepresentourworkonhoaxdetectiononacollectionofTweets.Wetacklethehoaxdetectionasatextclassificationproblem,withDoc2VecasthetextrepresentationmethodandSVMastheclassifier.Wecollectedandannotated5000Tweetsthatconsistof2500hoaxTweetsand2500truthTweets.Theexperimentalresultsshowthat theaccuracyofourproposedhoaxdetectiononTweetsis93.4%.
ElectronicNoseDatasetforClassifyingRiceQualityusingNeuralNetwork
FerdyErlangga,DedyRahmanWijaya,WawaWikusnaRice is a staple food ingredient because it is themain food element for Indonesia and theworld.However,thequalityofricecandeclineovertimeuntilitbecomesexpiredorsmellyandcannotbeconsumed.Atpresent,theconventionalmethodtodistinguishbetweenexpiredriceandnotexpiredriceisstillcarriedoutbyobservingricewiththehumansenseofsmell.However,thismethodisstillconsideredineffectivebecausethehumansenseofsmellcanchangeduetochangesinbodyhealth.Inthis case, this studyusesanelectronicnose (e-nose) andamachine learningneuralnetwork (NN)algorithmtodetectriceconsistency(expiredandnon-expired).Thedatasetwasobtainedfromthee-nosebyrecordingsensorinformationfor25weeksbystoring48.486totaldataand2.017datarecordsforoneweek.The results of the classificationusingNNarewith anaccuracy scoreof99.84%, theproposedmethodsuccessfullyclassifiedricequality.
SVMParallelConceptTestwithSMODecompositiononCancerMicroarrayDataset
RahmatRamadanPrasojoe,Setyorini
Support Vector Machine (SVM) is a reliable method for performing classification and regressionespeciallyinsupervisedmachinelearning.However,SVMhasscalabilityissuesincomputetimeandmemoryusage.Therefore,therearemanyproposalsforParallelSupportVectorMachine(PSVM)formining large-scale data. In this study, the authors conducted a PSVM concept test with SMOdecompositionthatcouldbehandledandclassifiedcancerusingmicroarraydata.Theauthorappliesthe Sequential Minimal Optimization (SMO) technique which uses LaGrange multipliers to solvequadratic programming (QP) problems that arise during training. To test the concept of SMOdecomposition,thedatasetwillbebrokendownintoseveralsubsetsandthenindependentlyconductSMO training foreach subset andcombineeach training result intooneSMOclassificationmodel.Evaluation is done by comparing the accuracy and performance of SMO decomposition and non-decomposition SMO. Evaluation result are accuracy of SMO decomposition 75% and non-decompositionSMO63%,andaswellasSMOdecompositiontrainingtime5.7timesfasterthantonon-decompositionSMO.
DetectingOnlineRecruitmentFraudUsingMachineLearning
HriditaTabassum,GitanjaliGhosh,AfraAtika,AmitabhaChakrabarty
ADVANCED PROGRAM | ICOICT 2021
90
OnlineRecruitmentfraud(ORF)isbecominganimportantissueinthecyber-crimeregion.Companiesfinditeasiertohirepeoplewiththehelpoftheinternetratherthantheoldtraditionalway.Butithasgreatlyattractedscammers.Inthispaper,wehaveproposedasolutiononhowtodetectORF.Wehavepresentedourresultsbasedonthepreviousmodelandthemethodologies,tocreatetheORFdetectionmodelwherewehaveusedourowndataset.WehavecreatedourdatasetbasedontheBangladeshjobfield and by using a publicly accessible dataset as a reference. Furthermore, Logistic Regression,AdaBoost, Decision Tree Classifier, Random Forest Classifier, Voting Classifier, LightGBM, GradientBoostingarethealgorithmsthathavebeenused.Wehavefoundtheaccuracyofdifferentpredictionmodels, where LightGBM (95.17%) and Gradient Boosting (95.17%) give the highest accuracy.Throughthispaper,wetriedtocreateaprecisewayfordetectingfraudulenthiringposts.
DataMiningforRevealingRelationshipbetweenGoogleCommunityMobilityandMacro-economicIndicators
Gunawan
Googlecommunitymobilityreportshavehelpedtoevaluatetheeffectivenessofgovernment-imposedmovement control among countries. However, the relationship between themobility data and thecharacteristicsofregionsislessreported.ThisstudyaimstorevealhiddeninformationfromGooglecommunity mobility reports and relate them to all 34 Indonesian provinces' macro-economicindicators.ThissecondaryresearchimplementsadataminingapproachusingtheCRISP-DMprocessframeworkandKnimeAnalyticsPlatform.ThecommunitymobilitydataofresidenceandworkplacearecollectedasatimeseriescoveringFeb16,2020,toJan31,2021.Macro-economicindicatorsarecollected from thewebsite of the Indonesiannational statistics agency. The clusteringmethodhasgroupedprovincesintothreebasedontheirmobility.ThefindingsindicatetherelationshipbetweenmobilityfluctuationduringtheCOVID-19pandemicandmacro-economicindicators,namelyhumandevelopment indexand labor forceparticipationrate. In the theoreticalaspect, thisstudyhasbeeninitiatingtheinvestigationofcommunitymobilityandmacro-economic.Policymakersindealingwithpost-pandemicrecoveryplanningmightconsidertheclustercharacteristicsforbetterplanning.
ADVANCED PROGRAM | ICOICT 2021
91
ABSTRACTSESSION4D:E-LearningandHCI
SuitableKnowledgeManagementProcessImplementation:ACaseStudyofPTSinergiSentra
Digital
YusufPratama,DanaI.Sensuse,SofianLusa,DamayantiElisabeth,NadyaSafitri,GhanimKanugrahan,BryanzaNovirahman
Somecompaniesnowadaysthatarenotfocusedoninformativetechnology(IT)developmenthavenotyet implementedaknowledgemanagement (KM)processwhich canbe seen from the lackofKMsystemfromtheapplicationthattheyhave.ThiswillthenleadtotheITdivisioninthecompanyhavingtodealwithstruggleinextracting,reading,aswellasreviewingmanykindsofdocuments.PTXYZ,oralso known as Port Cities Indonesia, is one of the company exampleswhichmainly focused theirbusinessondevelopingopen-sourceenterpriseresourceplanning(ERP)fortheothercompanies.Itisexpectedthatimplementingoneoftheperfectknowledgemanagementprocessesthatfitwiththecompanycansupporttheuseofmaintainingtheirknowledgeduetosomeintegrationtoitsdata.ThecontingencyfactortheorywasusedinthisstudytoobtaintheKMprocessneeds.Basedontheresultsof interviews with employees from each division involved, this study finds that Socialization forKnowledgeSharingisthemostprioritizedcontingencyprocesswithchattingandsharingmediatomanifestthatprocess.
CriticalSuccessFactorsforProjectTrackingSoftwareImplementation:ACaseStudyataBankingCompanyinIndonesia
HendroPrabowoHadi,RidhaEryadi,TeguhRaharjo
Intoday'sbusinessenvironment,locationisnotanobstacleanymore.Teamscanworkvirtuallyandcollaboratively with ease using web-based project tracking software. As a result of computertechnologyandinformationflowadvances,companiesaremovingtonewprojectmanagementtools,improvedprojectgovernance,andincreasedstakeholderengagement.Tokeepitscompetitivenessandabilitytodeliverproductsonschedule,oneoftheBankinIndonesiaadoptedJiraAtlassianasprojecttrackingsoftwaretoreplacethepreviousin-housesystemwhichiscallede-SDLC(electronicSoftwareDevelopmentLifeCycle) thatcanmonitorandtrackprojectstatus.But itdidnotgowellandfacedobstaclesatthebeginningofimplementation.ThisstudyaimstofindandevaluatetheCriticalSuccessFactors (CSF) for project tracking software implementation in one of Indonesia banking company.From previous research, there are identified challenges and obstacles when applying projectmanagement tools which are governance, adoption process, technology & knowledge, andstakeholders.Fromthisresearch,weidentified4keyscriteria,namelypeople,process,organization,and technicalasCSF (ordered fromthemost importantcriteriausinganalyticalhierarchyprocess-AHP).Thefindingsofthisstudycouldhelpcompanyincreasingthesuccessprobabilityinimplementingprojecttrackingsoftware.
AssuranceCasePatternusingSACMNotation
NungkiSelviandro
TheStructuredAssuranceCaseMetamodel(SACM)isametamodelandspecificationthatcanbeusedtorepresentstructuredassurancecases.Anassurancecaseisanapproachforanalysing,documenting,and communicating a clear structured argument and evidence about a particular systemwithin aspecific environment and circumstances. SACM provides abstract syntaxwith a set of features todevelop assurance cases, including supporting the development of an assurance case pattern. A
ADVANCED PROGRAM | ICOICT 2021
92
pattern in the assurance case development context is useful, for example, as an approach forabstractingthedetailsoftheargument,andwhenpossible,itcanbeusedforthedevelopmentofotherargumentsbyinstantiatingthepatterninaspecificdomainapplication.TosupportthedevelopmentandadoptionofSACM,wehavedevelopedSACMNotation(SACMN)asaconcretesyntaxthatconsistsofvisualvocabulariesandcompositionalrules(asvisualgrammar).ThedevelopednotationhasbeenincludedaspartoftheSACMstandard2.1updateversion.Inthispaper,weintroduceanddiscusstheapplication of the assurance case pattern using SACMN to support the SACM adoption in thedevelopmentofassurancecases.
SustainabilityAndAptnessOfGameElementsInAGamifiedLearningEnvironment
MageswaranSanmugamSustainingtheinterestlevelofstudentsinanyformoftechnology-basedlearningposesasignificantchallenge. Infusion of any tech-based elements into learningmay not be sufficient enough for theGeneration-Zstudents,whoselivesrevolvearoundtechnology.Althoughgamification,gameelementsinnon-gamingcontextsaremorecomfortablebeingimplementedbyordinarypeoplewithnotech-based knowledge.Nevertheless, in the context of learning, the sustainability of the game elementsneedstobeidentified.Therefore,mixed-methodresearchwasconductedon28studentsaged13yearsoldfromanurbanschoolinMalaysiafor14weeks.Thestudentswereintroducedtoagamifiedlearningmethodthatinfusedgamificationelementsinthetraditionalclassroomandtheonlineclassroom.Thegame elements testedwere points, badges, and leader boards. Studentswere taught two separatetopics in theMalaysian Science syllabususing gamified learning to ensure the students' continuityeffects.Uponcompletion,thegameelements'finaltallywasassessedandsupportedbytheinterviewfeedbackfromthetop3studentsfromeachgameelement.Basedonthefindings,gamifiedlearningwithgameelementshelpedreducedboredom,andusingtechnologymadelearningfun.Althoughsomerespondentssharedthefearofcomplacencyofusinggamesinlearningandthemixedresponsewasreportedwhenitcametothepreferredtypeofgameelementsortypeoflearningthatsuitsthegameelements.
UserInterfaceModelforVisualizationofLearningMaterialsinComicStripFormUsingGoal-
DirectedDesignMethod
MuhammadFauzanNurAdillah,DanangJunaedi,YanuarRahmanInalearningprocesstohelpstudentsrememberandunderstandthematerial,oneeffortcanbemadeistoprovideessentialpointsfromamaterial.Studentsarelessinterestedindeliveringessentialpointsbasedontextbecausetheyareconsideredmonotonous.Nowadays,manystudentstendtoreadcomicbooksratherthanbooks.Comicscanwashawaytheemotionsofreadersinthestoryline.Therefore,weneedsupportingmediatohelpteachersvisualizeessentialpointsinthebookincomicform.Sothattohelpteachersinthisregard,auserinterfacemodelisneeded.ThisresearchwasconductedwithaGoal-DirectedDesignapproachtogainuserperspectiveindevelopingsupportingtoolstovisualizethematerial'scrucialpointsintheformofcomicstrips.Testingtheresultingmodelwascarriedouttwiceandeachtestinvolved15respondentstogetmaximumresults.BasedontheevaluationresultsusingtheUSEQuestionnaireevaluationmethodfortheprototypemadeandtheresultswereintheexcellentcategory.Thus,theresultinguserinterfacemodelisasupportingmediumforteachersindeliveringlearningmaterialincomicformfollowingtheexpectedobjectives.
ADVANCED PROGRAM | ICOICT 2021
93
ABSTRACTSESSION4E:Networking,IoT,andSecurity
AccessibilityandResponseTimeAnalysisontheCOVID19WebsiteinIndonesia
RyanWicaksonoandHilalH.Nuha
Ping is a computer network utility software used to test the reliability of hosts on IP (InternetProtocol).Thisisusedtomeasurethetimesentfromthehosttothedestinationcomputerorcentralcomputer.PingoperatesbysendingtheInternetControlMessageProtocol(ICMP).Tracerouteisusedtoshowtheroutethepacketpassestoreachthedestinationorroutediagnostictool.Thisprocessisdone by sending the message Internet Control Message Protocol (ICMP) Echo Request_ to thedestinationwiththevalueofTimetoLiveorcommonlyreferredtoasTTLwhichisincreasing.Therouteshownisa listofrouter interfaces foundonthepathbetweenthehostanddestination.Thepurpose of pinging and traceroute is to find out the status of up or down in the network. Withtraceroute,wecananalyzeinformationaboutthelocationoftherouter.
ModifiedBitParityTechniqueforErrorDetectionof8BitData
FakhiraZulfira,HilalH.Nuha,DodiWisaksonoSudiharto,RioGunturUtomo
Bitparityisoftenusedasanerrordetectiontechniqueinsendingdigitaldatathathasinterruptionduringtransmission.Errordetectiontechniquesallowcorrectingdatatobeerror-free.Researchthathasbeenusedusingbitparityhasbeendonebutonlydetectedevenparityandoddparitywheretheprocessisverysimplesothatmodificationscanbemadetodetectotherthanevenoroddparity.Inthisstudyamodificationofthebitparitymethodwasbuiltinwhicheachmessagereceivedinthefirst4bitswillbedetectedwhether thebitvalue is amultipleof4and the remainingbitswillbeevenlydetected.Errordetectionusingtheproposedparitybitsresultsinafairlygooddetectionsystemandismoreefficientiftherearemorebitsinthemessage.IoTApplicationonAgriculturalAreaSurveillanceandRemote-controlledIrrigationSystems
RatnasariNurRohmah
This research applies IoT technology to help farmers overcome two problems by proposingagriculturallandsurveillancesystemandremote-controlledirrigationsystems.Theagriculturallandsurveillancesysteminthisresearchisdesignedtodetectobject'smotion,takepicturesofobjectsandsendimagedatatotheuser'ssmartphone.Thissurveillancesystemalsoprovidesa livestreamingvideomodebyrequest.Theirrigationsystemisdesignedtomonitortemperature,senddatatouser,andallowing to remotecontrol submarinepumpoperationby theuserviaa smartphone.Testonsystemsperformanceshowedthatsystemperformedproperly.Inthesurveillancesystem,theoptimaldistanceformotiondetectionbythesensorisin6meters.Onsunshineday,timeintervalfromimagetakenwhen somemotion detected and notification received by the userwas 4.9 seconds, whilstinterval time fromnotification to image sentwas3.9 seconds. In live streamingvideomode,usershouldwait2.3secondsinaveragetoreceivelivestreamingvideoonsmartphone.Intheirrigationsystemperformance,thesensormeasuredaveragetemperaturemeasurementerroris1.49%.Sensordata transmission to user's smartphone works well and user can remotely control the pumpoperation.
ADVANCED PROGRAM | ICOICT 2021
94
Present-80EncryptionAlgorithmImplementationonGPRSArduinoMega-2560CyberPhysical
TrackingSystem
DwiNovazrianto,RiniWisnuWardhani,NaufalHafizSyahidanCyberPhysical System(CPS) canbe characterizedasaphysical systemandanengineering systemwhich can be monitored, controlled, coordinated and integrated processes with computing andcommunication. Themore development ofmobile internet technology, themore vulnerabilities ininformation security. In the information age, cyberattacks are a challenge. Increasing security onCommunicationdevicecanmaptheactuallocationoftheGPSintoadifferentrepresentation.VTSwillsendvehicle locationdata to theserverorsystemuserperiodically.TheVTSdeviceconsistsof theArduinoMega-2560andthePRESENT-80algorithm.TheGPSmoduleservestoprovidelocationdata,and the sim800L v2 module as a communication module with GPRS services. PRESENT-80implementationontheVehicletrackingsystemwassuccessfullycarriedout.LocationdatacanbesentinciphertextformandlocationdatacanbedecryptedinanAndroid-basedmonitoringapplication.
HuntingCyberThreatsintheEnterpriseUsingNetworkDefenseLog
ArdianOktadika,CharlesLim,KalpinErlanggaAs business continues to evolve, the systems that support its functionalities also becomes morecomplex and sophisticated. This situation constitutes even larger cyber security threats, henceincreasedbusinessrisksthatrequiresmoresecurityprotectionforthebusinesssystems.Firewalls,IntrusionDetectionSystemandWebApplicationFirewallshavebeenutilizedextensivelytocounterthesethreatsattheperimeterbetweentheInternetandinternalsystems.Unfortunately,thesesecurityperimeter devices failed to detect new unknown threats due to its reliance on signature-basedmethods.Inthisresearch,weproposeaframeworkthatemploysthreathuntingthatutilizethesecuritydevice log and network forensics to uncover the unknown threats and validate the results usingmemoryforensics.Newrulescreatedfromthisdiscoverywillallowtheprotectionatthebordertobemoredynamicandeffectiveinhandlingpreviouslyunknownsecuritythreats.
ADVANCED PROGRAM | ICOICT 2021
95
ABSTRACTSESSION5A:Networking,IoT,andSecurity
BuildinganIDCardRepositorywithProgressiveWebApplicationtoMitigateFraudbasedon
theTwelve-FactorAppMethodology
KevinAkbarAdhiguna,FirhanMaulanaRusli,HendyIrawan
Dataandinformationhavebeenapartofhumanlife.TheyaremostlyaccessedthroughtheInternetandcanbeprovidedinvariouswayssuchastexts, files,andimages.Nowadays,a lotofpeopleusepersonalcomputersandmobilephonesandrelymoreoncloudcomputingservicestostore,backup,and share data. To cut down the probability of potential fraud, ID Card of users are stored in arepositoryinordertobeverified.TheIDCardrepositoryismeanttobeusedbothonmobilephonesanddesktopcomputers,sotheconceptofprogressivewebapplication(PWA)isutilized.Therefore,the ID cardPWA should fulfill cloud-native standards in order to gainmaximumbenefits of cloudcomputing. In this paper,wewill discuss a recommendation on how to develop a PWA using theTwelve-Factor App methodology and deploy it to a cloud infrastructure with cloud-relatedtechnologiessuchasDockerandAmazonWebServices(AWS).
XB-Pot:RevealingHoneypot-basedAttacker'sBehaviors
RyandyDjap,CharlesLim,AndiYusuf,KalpinErlanggaThedevelopmentofDeceptiontechnologysuchasthehoneypotispredictedtocontinuetotakeasharpdip.Wherethismakesperfectsenseduetothelargevolumeofattacks,complexityanddefenseevasionmanoeuvres. This causes the use of honeypot to becomemore prevalent in addition to deceptingserviceswhicharepopularlyattackedbyattackers.Honeypotsarealsousedtostudymethodsattackersusetocompromisecertainservices.Itishopedthattheaimoftheattackercanbeknownandafterthemappingcanbesearchedforattackpatternsthatcanbeusedasaruleto lookforpatternsthatareterminatedwhen the patterns found are similar to those in Honeypot. This research aims to takeadvantageofthehoneypotintermsofmappingtheattacksthatthehoneypothascollectedinordertoknowthetechnique,tacticandintentionofeachcommandthatislaunchedagainstthehoneypot.Apartfrommappingcommands.Thepurposeofthispaperistoprovideinsighttoreadersaboutthethreatsthat are rife with being modeled with statistical explanations using high volume and long tail. Inaddition, this research is a continuation of our previous paper where this research contained anenhancementframeworkandaddedhoneypotdatatypestounderstandbroaderattacks.DesignofaSnort-basedIDSontheRaspberryPi3ModelB+ApplyingTaZmenSnifferProtocol
andLogAlertIntegrityAssurancewithSHA-3
GarandYudhaandRiniWisnuWardhaniTheCOVID-19pandemichasforcedmanypeopletodoworkfromhome(WFH).Inpractice,networksusedinhomesgenerallydonotapplysecurityandlackawarenessofpossiblecybersecuritythreats.The development and use of security systems on home networks with inexpensive and practicalsolutionsarehighlydemanded.Therefore,adevicethatcandetectcybersecuritythreatsisneededtominimizetheriskswithpracticalsolutionsandaffordablecosts.ThisconductedresearchprovidesadesignofaSnort-basedIntrusionDetectionSystem(IDS)devicethatwasappliedtotheRaspberryPi3ModelB+.TaZmenSnifferProtocol(TZSP)isalsoimplementedtoanalyzenetworktrafficandTheSHA3algorithmusedtocalculateperiodicalhashvalue.Thispaperimplementsfivetypesofattacks,thereare ICMPBlackNurse, SYNFlood, SMTPBruteForce,RDPBruteForce, andWebPhishing. Inaddition, this research using the 'htop' program to perform performance testing, and 'sha3sum'
ADVANCED PROGRAM | ICOICT 2021
96
programtoconductthetestvectoralgorithm.Thesystemhas100%accuracyratedetectingtheattacksandrequireslessthan50%CPUand10%RAM.Sothatthesystemcanbeappliedtohomenetworksasapracticalandaffordablecostsolutioninordertoimplementcybercrimesecurity.
LearningMethodofPerformance-orientedCongestionControl(PCC)forVideoStreamingAnalysis
RezyNoerdyah,HilalH.Nuha,SidikPrabowo
TransmissionControlProtocol(TCP)congestioncontrolarchitecturesuffersfromperformanceissuesthat are not optimal. Thus making the TCP and the variations have little hope of achieving highperformance.ThisisduetotheuseofhardwiredTCPmappingwhereeacheventhasbeenassumedtobeaspecificdisruptionandTCPmustcopewiththeincidentwithoutunderstandingtherealconditionofthenetwork.Thisassumptionresultsinperformancedegradation.Performance-orientedCongestionControl(PCC)isanewcongestioncontrolthatmakeseverysenderobservingtheactionandnetworkperformanceempiricallytobeabletotakeactionthatyieldshighperformance.PCChasbeentestedinseveralcases.Onecaseisvideostreaming.Theexperimentisdesignedtoobtaintheperformanceforvideostreamingintermsofthroughput,delayandpacketlossforthePCCandtheTCPtodeterminebetterperformanceresults.ResultsfromthetestingofeachmetricinwhichTCPandPCCthroughputsare 1064.841 and 150.825 kbps respectively. Delay of TCP and PCC are 5.326 ms and 3.843 msrespectively.PacketlossofTCPandPCCare0.905%and0.016%,respectively.SothePCCachievesgoodperformancesontheparametersofdelayandpacketloss.WhereastheTCPisshowntoperformbetterintermsofthroughput.
ExperimentalInvestigationofWaveAbsorberMadeofRingResonator-BasedAMCStructure
LevyOliviaNur,IchsanNusobri,BudiSyihabuddin,AchmadMunir
Thispaperdealswithanexperimentalapproachforinvestigatingcharacteristicsofanelectromagnetic(EM)waveabsorbermadeofartificialmagneticconductor(AMC)structure.AnarrayofringresonatorsinsymmetricalshapeusedasthebasisofAMCstructureisdesignedtoabsorbtheincomingEMwavesatacertainoperatingfrequency.Theproposeddesignofwaveabsorbermadeofringresonator-basedAMCstructureisdeployedonaFlameRetardant(FR)4Epoxydielectricsubstratewiththethicknessof1.6mm.Theshapeanddimensionof ringresonatorareoptimized toworkat thecertainoperatingfrequencyusingaunit cellwithproperboundaryconditions.To investigate theabsorptivity rateofproposed wave absorber, resistive elements are incorporated into the ring resonator. Here, theexperimentalmeasurementaimstoverifytherealizedwaveabsorberwiththeproposeddesignandtomeasure the absorptivity characteristics of realized wave absorber. The measured results for therealizedwaveabsorberwhichtakesthedimensionof220mmx220mmhavetheabsorptivityrate,intermsofreflectioncoefficient(S11),of-16.65dBand-25.72dBforthewaveabsorberwithoutandwithresistors,respectively.TheseresultsarecomparabletothesimulatedoneswiththeS11valuesof-18.01dBand-23.41dBforthedesignwithoutandwithresistors,respectively.
ADVANCED PROGRAM | ICOICT 2021
97
ABSTRACTSESSION5B:DataScience
InformationCascadeMechanismandMeasurementofIndonesianFakeNews
AndryAlamsyahandAslaSonia
Thehighnumberofsocialmediaactorshasthepotentialtoproducefakenews.Fakenewscouldbemotivatedbyvariousagendas,suchaspolitics,government,andhealth.Therefore,weneedtoknowhow themechanism andmeasurement of the spread of fake news.One approach to studying thespreadof it is the informationcascade.Thispaperwillmodel the informationcascademechanismusing Social Network Analysis (SNA) and Susceptible-Infected (SI) model. SNA is adopted toinvestigatethespreadingmechanismanddeterminetheproportionofactorsexposedtofakenews.SIisappliedtomeasurethespeedoftransmissionoffakeandtruenews.Usingseveraltopicsamples,theresultsallowustounderstandthemappingofcascadeinformationfromfakenewsbyalevelthatdifferentiatesthenodelevelfromthesourcenewstotherestofthenodesinthenetwork.Ourfindingfakenewscanreach0,6414morefractionsandspread4,6timesfasterthantruenews.
FraudAccountsIdentificationModellingonMulti-PlatformE-Commerce
GrawasSugihartoandYudistiraAsnarNowadays,cybercrimeisincreasinglyprevalentinsociety.BasedondatacompiledbytheIndonesiaNationalPolice,thenumberofcybercrimesincreasesby6.46%annually,withonlinefraudasthemostreportedcrimewith7.892casesor44.40%outofthetotalcaseshandled.Themodusoperandiinonlinefraudprimarilyusesmanipulatedprofileaccounttogainthevictims'trust.Therefore,itisnecessarytohavea commondetectionmodel for fraudaccountsonmulti-platforme-commerce toavoidonlinefraud. This research uses theNaïve Bayes classification, Decision Tree, and K-NN as themodelingalgorithms. The classification test result showed that the optimal performance with the highestaccuracydiffersamongtheplatform.ThegreenplatformreachesthehighestaccuracyusingtheK-NNalgorithm (90.51%), the red platform went to the optimal performance using the Decision Treealgorithm(96.89%),andthemulti-platformreachedtheoptimalperformanceusingtheNaïveBayesalgorithm(90.02%).ClassificationonParticipantsRenewalProcessinInsuranceCompany:CaseStudyPTXYZ
DeddyUtomo,NoperidaDamanik,IndraBudi
Insuranceisaformofriskmanagementandoneofthefastest-growingbusiness.PTXYZisacompanythatfocusesonhealthandlifeinsurance.OneexcellentproductownedbyPTXYZisManagedCare(MC)Insuranceanditdominate64.5%ofthecompany'spremiumincome.However,MChasahighclaimratiovalue.Provenbytherewere363companiesthathaveaclaimratioofmorethan76%in2020.Theincreasingoftotalclaimratioisduetothecompanyhasnotbeenabletopredicttheclaimratiowhentherenewalcompanyapplyforaninsuranceparticipant.Thisstudyfocusesonclassifyingparticipantson insurance renewalprocess so that company canbemore selective to approve theparticipants.Participantselectioncanhelpcompanytoreduceclaimratio.Proposedmethodisdoingclassificationoninsuranceparticipantsdatausing3803datasetswith4attributesandfivealgorithmsandfindsignificantfeatureswhengeneratingthemodel.Themodelswillbevalidatedusingk-foldscross-validationwithk=10,evaluationresultsshowtheaccuracyofeachalgorithmasfollowing,NaïveBayes70.00%,SupportVectorMachines67.00%,DecisionTree95.40%,LogisticRegression90.20%,andNeuralNetworks79.30%.TheresultsofthestudyrecommendDecisionTreealgorithmwithanaccuracyof95.40%fortheclassificationofrenewalcompanythatwilljoinasinsuranceparticipants
ADVANCED PROGRAM | ICOICT 2021
98
because it has a better accuracy value than other algorithms. Decision Tree shows that themostsignificantfeaturesindefiningprospectivecompanyassessmentistheaverageofage.HybridSpace-TimeModelandMachineLearningforForecastingMultivariateSpatio-Temporal
Data
HendriPrabowo,DedyDwiPrastyo,SetiawanThepurposeofthisstudyistoproposeahybridmodelbycombiningstatisticalmethods,namelyTimeSeriesRegression(TSR),MultivariateGeneralizedSpace-TimeAutoregressive(MGSTAR)asaspace-timemodel,andMachineLearning(ML)toforecastmultivariateSpatio-temporaldatasimultaneously.Thelinearmodel,namelyTSRisusedtocapturetrendsanddoubleseasonalpatterns.MGSTARisamodelforcapturingdependenciesbetweenlocations.Meanwhile,capturingnonlinearpatternsusedtheMLmodel. In this study, three types ofMLmodel is used, i.e., Deep LearningNeural Network(DLNN),FeedForwardNeuralNetwork(FFNN),andLongShort-TermMemory(LSTM).Weapplythisproposedmethodtosimulateddata.BasedontheRootMeanSquareError(RMSE)value,theproposedhybrid methods, namely TSR-MGSTAR-DLNN, TSR-MGSTAR-FFNN, and TSR-MGSTAR-LSTM,outperformothermodelssuchasTSR,MGSTAR,MGSTAR.-DLNN,MGSTAR-FFNN,MGSTAR-LSTM,andTSR-MGSTAR,especiallywhenthedatacontainnonlinearnoisecomponents.TheresultsalsoshowthattheproposedhybridmodelcantacklecomplexpatternsonSpatio-temporaldatacontainingtrends,doubleseasonal,linearnoise,nonlinearnoise,anddependenciesbetweenlocations.
ComparativeStudyofCovid-19TweetsSentimentClassificationMethods
UntariN.Wisesty,RitaRismala,WiraMunggana,AyuPurwariantiCovid-19 isadiseasecausedbyavirusandhasbecomeapandemic inmanycountriesaround theworld.Thediseasenotonlyaffectspublichealth,butalsoaffectsotheraspectsoflife.Peopletendtowritecommentsaboutthingshappeningduringthepandemiconsocialmedia,oneofwhichisTwitter.SentimentanalysisonTwitterdataisnotaneasytaskduetothecharacteristicsofthetweetertextwhichisusergeneratedcontent.Therefore,inthispaper,asentimentanalysisstudyiscarriedoutonTwitterdatausing threeschemes,namely thevectorspacemodel (BagofWordsandTF-IDF)withSupportVectorMachine,wordembedding(word2vecandGlove)withLongShort-TermMemory,andBERT (Bidirectional Encoder Representations from Transformers). Based on the conductedexperiments,BERTachievedthebestperformancecomparedtotheothertwoschemes,reaching0.85(weightedF1-score) and0.83 (macroF1-score) for the classificationof three sentiment classes onKagglecompetitiondata(CoronavirustweetsNLP-TextClassification).
ADVANCED PROGRAM | ICOICT 2021
99
ABSTRACTSESSION5C:DataScience
CountDataForecastingusingPoissonAutoregressionforCOVID-19CasePredictioninJakarta
BahrulIlmiNasution,YudhistiraNugraha,JuanKanggrawan,AlexLukmantoSuherman
COVID-19iscurrentlybecometheglobalproblem,includinginJakarta,Indonesia.TherehavebeenmanyapproachestopredictingtheCOVID-19occurrence.Oneoftheoccurrencesisusingforecastingapproach.Thetraditionalforecastingmethod,particularlymachinelearning,oftendoesnotconsidertheconditionofthedata,althoughithasformsofthecount,suchasthenumberofcases.ThisstudyaimstoemployanautoregressionmodelusingPoissondistributioninpredictingtheCOVID-19futurecases, namely the positive and recovery number. We compare the Poisson Autoregression withseveralwell-knownforecastingmethods.ThisstudyfoundthatPoissonAutoregressioncouldcreateanaccuratepredictionandtendtofollowstherealdataforthenext8to14daystothefuture.ThisapproachcanbeusedtoforecastthefuturecasesofCOVID-19,andothercasesthatusecountdata.OptimizationofCropsAllocationPlanninginCianjurInvolvingWaterCostConstraintsUsing
GeneticAlgorithm
BambangWahyudi,IrmaPalupi,SitiSa'adahTo become urban agriculture, Cianjur needs to create land-use planning base on its agriculturalsituation.Inthiswork,anoptimizationmodeltocomputeoptimalcropallocationsforeachsub-districtis proposed and solved by a genetic algorithm (GA) to find the best solutions. The land utility isformulatedasthemarginbetweenthepossiblecostandrevenuebaseonthedatacollectedfromtheIndonesian Central Bureau of Statistics data. The result analysis can support decision-making forallocationcropproductionplanninginpracticetoincreaseagriculturalrevenueintheCianjurregency.Ageneticalgorithmisusedtofindtheoptimalsolutionthatmaximizesthemarginfunctionwithsomerealisticconditionsobtainedfromthegovernment'sallowedlandrestriction.Thecomputationresultsshowthatthehighestoptimalvalueis30.65onthelogscale.Thefitnessconstanttothisasymptoticriseisafter17generations,sothatitconvergesquicklyforsomeexperimentalGAparametersetting.Intheoptimalallocationresults,thesweetpotatogainsthesmallestportion,followedbycassavathatis only around2%of the total plantation area.Meanwhile, the largestportionof landallocation issuggestedtobecorn,basedontheoptimalsolutionreturningthemaximumprofitmargin.Forallcrops,theCidaunsub-districtappearstobethemostexpansivelandareainCianjur.
FakeNewsDetectionwithHybridCNN-LSTM
KianLongTan,ChinPooLee,KianMingLim
In thepastdecades, informationandcommunicationtechnologyhasdevelopedrapidly.Therefore,socialmediahasbecome themainplatform forpeople toshareandspread information toothers.Althoughsocialmediahasbroughtalotofconveniencetopeople,fakenewsalsospreadmorerapidlythanbefore.Thissituationhasbroughtadestructiveimpacttopeople.Inviewofthis,weproposeahybridmodelofConvolutionalNeuralNetwork(CNN)andLongShort-TermMemory(LSTM)forfakenews detection. The CNN model plays the role of extracting representative high-level sequencefeatureswhereastheLSTMmodelencodesthelong-termdependenciesofthesequencefeatures.Tworegularizationtechniquesareappliedtoreducethemodelcomplexityandtomitigatetheoverfittingproblem.TheempiricalresultsdemonstratethattheproposedCNN-LSTMmodelyieldsthehighestF1-scoreonfourfakenewsdatasets.
ADVANCED PROGRAM | ICOICT 2021
100
AspectBasedSentimentAnalysisWithCombinationFeatureExtractionLDAandWord2vec
RizkaVioOctrianyInggitSudiro,SriSuryani,YuliantSibaroni
Aproductreviewisneededbyacustomerbeforehebuysaproduct.Currently,severalplatformscanbeused toprovideproduct reviews, oneofwhich is thebeautyproduct. Every customer can readbeautyproductreviews,notonlyfromoneaspectofthereviewbutitcanbefromseveralaspectsofthereview.itisdifficultforconsumerstofindallthereviewsfromvariousaspectsquickly.Therefore,inthisstudy,acombinationofLDAmodelingmethodsandWordEmbeddingWord2vecwereused,toobtainsentimentsfromeachofthepredeterminedaspectsofthereview.Inthisstudy,theaccuracyofthecombinationofLDAwillbecomparedwiththeWord2vecSkip-gramandContinuous-bag-of-word(CBOW)models.Fromthe twocombinations, it is foundthat thecombinationaccuracyofLDAandWord2vecSkipgramis80.36%,andforCBOWisonly74.37%.Meanwhile,theSVMandK-FoldCross-Validationalgorithmsareusedtofindtheaccuracyofsentimentpredictionsontheaspectsofprice,packaging,andfragrances.Comparedtotheothertwoaspects,thepackagingaspecthasthehighestaccuracyat89.71%.
SentimentAnalysisonBeautyProductReviewsusingLSTMMethod
MuhammadRafiiDanendra,YuliantSibaroniA review isanopinion that containsvalueon the joboreventbeing reviewed.Manysitesprovidereviews of products or goods in themodern era to users, such as the femaledaily.com site,whichprovidesaplatformforuserstoreviewproductspurchased.Thesentimentscontainedinthesereviewsarevaluableinformationforbusinessowners.Thankstoproductreviews,businessownersgetinsightsanddatarelatedtotheproductstheyselltoimprovetheirproducts'quality.However,gettingopinioninformationfromanunstructuredreviewtextisquitedifficult.Thisstudyaimstoclassifythesereviewsas"positive"or"negative".ThemodelproposedforclassificationisLSTM.LongShort-TermMemory(LSTM)wasusedinthepreviouslytrainedmodeltoclassifythisreview.Themodeldesignedforthemodel focuses on preprocessing reviews as follows: data cleansing, case folding, normalization,tokenization,stopword,andstemming.Onceclassified,thisreviewisvisualizedasagraph.Thebest-casescenarioresultswithanaccuracyof95,10%onthesentimenttowardsthepriceaspect.
ADVANCED PROGRAM | ICOICT 2021
101
ABSTRACTSESSION5D:DataScience
IndonesianIDCardExtractorUsingOpticalCharacterRecognitionandNaturalLanguagePost-
Processing
FirhanMaulanaRusli,KevinAkbarAdhiguna,HendyIrawan
Thedevelopmentofinformationtechnologyhasbeenincreasinglychangingthemeansofinformationexchangeleadingtotheneedofdigitizingprintdocuments.Inthepresentera,thereisalotoffraudthatoftenoccur.Forexampleisaccountfraud,toavoidaccountfraudtherewasverificationusingIDcardextractionusingOCRandNLP.OpticalCharacterRecognition(OCR)istechnologythatusedtogenerate text from image. With OCR we can extract Indonesian ID card or \emph{kartu tandapenduduk} (KTP) into text using 3 different OCR library, PyOCR, Pytesseract, and TesseOCR. ToimprovedtheaccuracywemadetextcorrectionusingNaturallanguageProcessing(NLP)basictoolsto fixing the text. With 50 Indonesian ID card image we compared the performance with threedifferentOCRlibrary.TheresultofourexperimentshowthatPytesseracthadthebestperformancewith0.78F-scoreand4510millisecondstoextractperIDcard.
AnalysisofRecordsManagementMaturityLevelforDataCollectionofNetworkAssetsinIndonesianTelecommunicationIndustry
RizkyA.C.EkaPutri,AchmadHidayanto
PT.XYZisastate-ownedcompanyintelecommunicationsandinformationtechnology.Indevelopingitsbusiness,PT.XYZhasservedmorethan2.200corporatecustomersand22.760retailcustomersbyhavingalotofnetworkassets.Thelargenumberofnetworkassetsownedmeansthatthecompanymustbeabletocarryoutgooddatacollectiontofindoutinformationonthedistributionandcapacityof assets that are still available. Themost commonproblems encountered at PT.XYZ is about datarecordsmanagementofnetworkassets.PT.XYZalreadyhasasystemfordatacollectionofnetworkassets, but it's still not running well. This occurs because there are no procedures and clarity ofaccountabilityrelatedtorecordsmanagement,sothereweremanydatacollectionprocessesthatwerestillmanually.Basedontheproblem,thepurposeofthisstudyistoconductanassessmentofrecordsmanagement level at PT.XYZ and provide recommendations to increase the level. The assessmentprocessandrecommendationswillbebasedonARMAGARPFramework.Thiswaschosenbecauseit'sa comprehensive approach that widely used in managing the information assets of companies ororganizations.Theresultsofmaturitylevelassessmentofthisstudyareatlevel2foraccountability,transparency, compliance, availability, retention, and also level 1 for integrity, protection, anddisposition.TheseresultsalsoindicatethatthecompanyneedstoimproveseveralprocessesintheGAPRPrinciples of Integrity, Protection, andDisposition because there is a large gap between thecurrentlevelandtheexpectedtargetlevel.
DataAcquisitionGuideforForestFireRiskModellinginMalaysia
YeeJianChew,ShihYinOoi,YingHanPang
Availabilityofremotesensingdata(i.e.,informationcapturedfromsatellite)inconjunctionwiththeusageofGeographicInformationSystem(GIS)hasmadeitfeasibletodeliverafiremodelcapabletosegregatetheareaintoahigherorlowerriskfireregion.Theadvancementoftechnologieshasalsoinauguratedthepossibilityto incorporateremotesensinginformationandothergrounddata(e.g.,meteorological data, distance to road data, etc.) by utilizingmachine learning classifiers or deeplearningalgorithmtopredicttheforest fireoccurrence.However, itshouldbehighlightedthatthe
ADVANCED PROGRAM | ICOICT 2021
102
dataacquisitionproceduremayvarydependingonthevicinityofthestudyareasincesomedataareonlyobtainablefromthespecificgovernmentauthority.Inthispaper,wewillbedisclosingsomeofthe publicly accessible remote sensing data and some of the valuable data attainable from theMalaysian government that is useful for detecting forest fire in Malaysia. Additionally, previousstudiesandworksthathaveemployedthedatasourcetomapforestfirearealsodeliberatedinthispaper.OnlythedatathathadbeenexploitedinthepastforMalaysiaarediscussed.
ImplementationofHiddenMarkovModel(HMM)toPredictFinancialMarketRegime
BambangWahyudi,IrmaPalupi,AgungPutraThisworkperformshowtoimplementtheconceptofHiddenMarkovModel(HMM)tofindfinancialmarkettrendforgivenonlytheobservedstateobtainedfromthestockprice.Theconsideredmarkettrendissetasahiddenstate, that inthe financial technicalanalysisknownasBearish,Bullish,andSideway,which are important for decisionmaking of stock trading in order to recognize the goodmomenttosell,tobuyortojustholdtheshares.InordertoobtainthemostlikelysequenceofhiddenstatesthroughHMM,whichiscomputationallycanbeadynamicprogrammingproblem,weexplainhow the Viterbi algorithm work for the case in this study. To get the stock price prediction asobservationstates,theARIMAmodelisusedbasedonexperimentaltrialoffittingmodel,thenusetheresultasapredictedobservedstatesthatbetheinputtopredictthemarkettrendusingHMMfortheshortperiodoffuturetime.Severalinterestingresultsofhiddenmarkettrendanditsstudyarealsoprovided,includingtheaccuracy,precision,recallandtheconsistencyofthemodeltothegivendataset.
PredictionofGraduationwithNaïveBayesAlgorithmandPrincipalComponentAnalysis(PCA)
onTimeSeriesData
WishnuDwiHerlambang,KusumaAyuLaksitowening,IbnuAsrorOn-timegraduationisanachievementforbothstudentsanduniversity.Foruniversity,especiallystudyprogram,thepercentageofgraduationon-timeisacrucialpointforaccreditation.Therefore,itrequiresagoodstrategy.Thepercentageofgraduationontimeincollegecanbepredictedwithdataminingandmachinelearning.Theobjectiveofthisresearchistoprovideearlierinformationregardingstudentswhoareatriskonnotgraduatingontime.Thus,thestudyprogramcantakeappropriateactionbeforeit is too late.Thereareseveralclassificationmethodsthatcanbeusedforprediction.OurresearchcombinesNaïveBayeswithPrincipalComponentAnalysis(PCA).PCAisusedtosimplifythecomplexacademic data. The PCA result that has simpler structure then to be processed usingNaive Bayesclassification.ThisresearchusesdataobtainedfromfourbatchofstudentacademicperformancedatainInformaticsStudyProgram,TelkomUniversity.Thedatasetispartitionedbyacademicyeartoobtaintimeseriesdataofeachstudent.ThecombinationofPCAandNaïveBayesalgorithmsobtainedbetterresultcomparedtoclassificationusingNaïveBayesonly,withinaverage3.65%higheraccuracy.
ADVANCED PROGRAM | ICOICT 2021
103
ABSTRACTSESSION5E:DataScience
ComparativeAnalysisofSupportVectorMachine(SVM)andRandomForest(RF)Classification
forCancerDetectionusingMicroarray
Irawansyah,Adiwijaya,WidiAstuti
Cancer is the second leading cause of death globally.According to theWorldHealthOrganization(WHO)in2018,approximately9.6milliondeathswerecausedbycancer.Globally,about1in6deathsarecausedbycancer.Onewaytodetectcanceristousemicroarraydataclassification.Microarraytechnologyisusedtodetecttheexpressionofthousandsofgenesatthesametimetoanalyzeanddiagnose cancer. However, microarray data has high dimensions because of its large number offeaturesandlowdatadistribution,whichmeansthatithasasmallnumberofdatasamples,whichcauseslowperformance.Toovercomethisproblem,dimensionreductionisneeded.Therefore,itisnecessarytoreducethedimensionsofmicroarraydatawithRandomProjection(RP)toreducethehighdimensionsandusetheSupportVectorMachine(SVM)andRandomForest(RF)asclassificationmethods.TheclassificationmethodwillbecomparedandanalyzedtodeterminewhichclassificationmethodproducesthebestperformancebyusingRandomProjection(RP)asadimensionalreductionmethod.Basedonthesystemthathasbeenbuilt,thebestaccuracyforcolontumordatais69.23%withRandomProjection (RP)-SVM,LungCancer is 100% forbothmethods classification,OvarianCancer is 100% for bothmethods classification, the prostate tumor is 95.12% for bothmethodsclassificationand66.66%forbothmethodsclassification.EvaluatingtheBPPTMedicalSpeechCorpusforAnASRMedicalRecordTranscriptionSystem
ElviraNurfadhilah,AsrilJarin,LylaRuslanaAini,SiskaPebiana,AgungSantosa,MuhammadTeduh
Uliniansyah,EduwardButarbutar,Desiani,Gunarso
In joint research fundedbyLPDP,BPPT is collaboratingwithSolusi247andHarapanKitaNationalHeart and Vascular Hospital to develop a speech corpus named BPPTMedical Speech Corpus forbuildingamedicalASR(automaticspeechrecognition)system.Thecorpuswasproducedbyrecording100 speakers for 81.68 hours. The corpus has 600 unique sentences containing medical terms.Previously,BPPThasdevelopedaspeechcorpusforthegeneralnamedBPPTGeneralSpeechCorpus.AthirdspeechcorpusnamedBPPTCombinedSpeechCorpuswascreatedbycombiningthemedicalandgeneralcorpus,andeachcorpuswastrainedusingPyChaintobuildanASRmodel.Basedontheexperimentsweconducted,theASRmodelbuiltfromthecombinedspeechcorpushasthebestWERof6.10%.
ImplementationofSimulatedAnnealing-SupportVectorMachineonQSARStudyofIndenopyrazoleDerivativeasAnti-CancerAgent
MuhammadFajarRizqi,RezaRendianSeptiawan,IsmanKurniawan
Cancerisadiseasethatoccursduetotheuncontrolledgrowthofabnormalcellscausingbodytissuedamage.Thisdiseaseisconsideredasadeadlydisease.In2019,1700deathsoccureverydayduetocancer.Someeffectiveanticanceragentsareknowntocausetemporarytochronictoxiceffects.Thereare several compounds that have the potential to become anticancer drugs, one of them isindenopyrazole.Thegoalof thisresearch is to implementsimulatedannealingandsupportvectormachinemethodintheQSARstudytopredicttheactivityofindenopyrazolederivativesasanticancerdrugs.Simulatedannealingisusedforfeatureselectionandsupportvectormachineisusedformodeldevelopment.Inthisresearch,weusedthreekernelmodelsforSVM,namelySVMwithRBFkernel,
ADVANCED PROGRAM | ICOICT 2021
104
SVMwithlinearkernel,andSVMwiththepolynomialkernel.Fromthreemodelsthatwereregressed,SVMwithRBFkernelhasparameterC=10,gamma=scaleandepsilon=0.1produceR2scoretrainandtest 0.79 and 0.60, respectively. SVM with linear kernel has parameter C=1000, degree=1 andepsilon=0.1produceR2scoretrainandtest0.61and0.63,respectively.SVMwithpolynomialkernelhas parameter C=1000, degree=2 and epsilon=0.1 produceR2 score train and test 0.72 and0.50,respectively.Basedonthevalidationresults,onlymodelwithRBFkernelwhichparameterssatisfyallthecriteria.FromtheresultwecanconcludethatthemodelwithRBFkernelisthebestmodelandacceptable.
RansomwareDetectiononBitcoinTransactionsUsingArtificialNeuralNetworkMethods
Hairil,NikenCahyani,HilalH.NuhaTheuseof digital currencyor cryptocurrency in various virtual transactions is commondue to itseasiness. Cryptocurrency is a digital currency that is used for virtual transactions on the internetnetwork.ThemostcommontypesofcryptocurrenciesincludeLitecoin,Ethereum,Monero,Ripple,andBitcoin.Eventhoughcryptocurrencieshavesecretcodesthatarequitecomplicatedandcomplexthatservetoprotectandmaintainthesecurityofdigitalcurrencies,itispossibletobehackedbyskilledhackers.Cryptocurrency-relatedhackingisatypeofdigitalcrimethatisveryharmfulanddangerous.Forexample,inrecentyears,casesofhackingonbitcointransactionsusingransomwarehavebeenontherise.Ransomwareismalicioussoftwarethatsecretlyinfectsavictim'sdevice,andsuddenlyasksforaransomtodecryptencrypteddata.Asthenameimplies,ransomwhichmeansransom,thistypeofmalwareaimstoblackmailavictimwhosecomputerisinfectedwithransomwarebyaskingforacertain amount ofmoney as a ransom.Therefore, a designwasbuilt in the formof a ransomwaredetection system based on available bitcoin heist data so as to minimize hacking attacks againstcryptocurrencyinthefuture.TheransomwaredetectionsystemwasbuiltusingthebackpropagationartificialneuralnetworkmethodusingWekasoftware.Thebestresultsindatatestingareusingtheparameternumberofhiddenlayerwith9neurons;learningrate0.1;andthenumberofiterationsof5000yieldsanaccuracyrateof97%.
EmotionalContextDetectiononConversationTextwithDeepLearningMethodUsingLong
Short-TermMemoryandAttentionNetworks
AfridaHelen,MiraSuryani,HidayatulFakhriTheconversationinthetextisaninterestingresearchonNaturalLanguageProcessing.Oneofthetextconversation tasks is to know the emotions of the people involved in the conversation. TheconversationinsocialmedialikeTwitter,Instagram,shortmessageservice,WhatsApp,andsoon,ofteninvolves emotion. Somehow the comments are impulsive sentences that can stimulate emotions.Expressing emotionsusing text is rarelydone anduncomfortable.However,withnatural languagetechnologydevelopment,expressingemotionsusingtextcanbesucceededwithspecificsymbols.Wecall thespecific symbolsemojis.Somanyemojis canexpressemotions.This researchproposes theemoji symbol as a character feature. We introduce the Emoji2Vec method and Long Short-termMemorywithAttention.TheAttentionthatisusedhasacomplextopology.Wecomparedtheresultsofthisstudywiththebaselinemodel.Themethodweproposeisbetterthanthebaselinemodel.
ADVANCED PROGRAM | ICOICT 2021
105