Post on 26-Jan-2015
description
transcript
Trends in Mobile Data Analytics
Satnam Singh, PhD
Senior Chief Engineer,
Samsung India Software Operations, Bangalore
[Thanks – Ravindra Guntur, Jithendra Vepa, Balvinder Singh
and other researchers mentioned in References]
2
Motivation: Context-Aware Intelligent Devices
SpeechMusicImage/
Video
Gesture
Location
Text
Acceleration
�Transform mobile devices into
intelligent context-aware systems
�Learn from user’s context,
history of interactions, and state
of the physical environment
�Transform mobile devices into
intelligent context-aware systems
�Learn from user’s context,
history of interactions, and state
of the physical environment
3
Speech Analytics: Trends
Siri [Apple]S-Voice [Samsung]
Google Voice &
Google Now
[Google]
Vlingo [Nuance]
• Current Virtual
Assistants (VAs):
– Local Search and maps
– Weather
– Voice dialing, SMS and email, etc.
• Future VAs - knowledge-
centric
• Understand user’s intent
[Beyond Siri]
Virtual Assistants Technology
4
• Future VAs need to perform well in noisy environment
Text
Text
5
Text Analytics: Trends
[SGI]- Twitter's 'heartbeat
[Socialbakers]
• B2B Applications: Brand building and Social
Marketing
• Social network-driven
Recommendation Engines
6
Text Analytics: Technology
Lexicon or
taxonomy or
Ontology
Term
Clustering
and
Classification
Unstructured
Text Corpus
Unstructured
Text Corpus
Lexicons or
Dictionaries or
Ontologies
Lexicons or
Dictionaries or
Ontologies
Text
Representation
• Beyond “searching”
• Enables What-If Analysis, Information Retrieval and
Detection
7
Image Analytics: Trends
• SaveUp -Visual Search:
[recognize.im]
• Vivino app- Wine Search [Kooaba]
• Viewing crowdsourced content
[CrowdOptic]
• Google Goggles, Nokia Point,
Amazon Snaptell, Ricoh iCandy,
Vivino
CrowdOptic
Saveup
Image Analytics: Technology
Object Recognition
Crowd Optic
Detect and Cluster Events
[B. Girod]
[A. Stefan]
9
Video Analytics: Trends and TechnologyVideo Motion Detection
• B2B Applications in Surveillance and
Remote Monitoring
• Video Enhancements
• Technology: Image and Object
Recognition, Machine Learning
Intrusion Detection
Video Panorama
[IntelliVision]
• Navigate a smart TV with
hand gestures [Tarsier]
• Hand gesture-based start
and stop music [Flutter]
• Goodbye to your Mouse
and Keyboard [Leap
Motion]
• Various Users: Surgeons,
Gamers, Artists, Engineers10
Gesture Analytics: Trends
FluttterFluttter
• Computer Vision - Template Matching use
Semaphores – sign language, Tracking and
morphing
11
Gesture Analytics: Technology
Feature Extraction
and hand detectionPattern
Matching with
Existing
Templates
(Hidden Markov
Models)
Gesture-
Driven
Control/Action
12
Multi-Model Analytics: Trends
• Indoor Maps – airports,
hospitals, etc.
• 3D Maps- Cool 3D Maps
[eeGeo]
• Activity Recognition: Detect
walking, driving, biking, climbing
stairs, standing, etc. [Alohar
mobile, ActiServ]
• Indoor Map Technology- Wi-Fi fingerprinting: Cisco Systems, Qualcomm and indoor map developer Meridian
• Indoor Map Startups – [Aisle411,Wifarer, Micello, Meridian, Point Inside and MapEverywhere]
• Activity Recognition Technology : A combination of Accelerometer/GPS, timing and wifi data
13
Multi-Model Analytics: Technology
[ActiServ]
• Mobile data analytics is still in infancy stage
• Mobile Data Analytics: – Would bring innovative features in next generation
smartphones (B2C Opportunities)
– Transform the businesses through their deep integration in B2B solutions
• Future mobile devices would be build on analytics platform and deliver intelligent, personalized, context-aware features and services
14
Summary
• Beyond Siri, http://www.visionmobile.com/blog/2012/06/infographic-beyond-siri-the-next-frontier-in-user-interfaces/
• Nuance Communications,
• https://www.recognize.im/site/showcaseApps
• B. Girod, “Mobile Visual Search,” IEEE Signal Processing Magazine, July 2011.
• http://www.usatoday.com/story/tech/2012/11/26/indoor-map-technology-poses-challenges-and-opportunities/1698739/
• Tarsier Inc. http://www.moveeye.info/
• LeapMotion, https://leapmotion.com/
• SGI -Global Twitter Heartbeat, http://www.sgi.com/go/twitter/
• Flutter, https://flutterapp.com/
• A. Stefan, V. Athitsos, J. Alon, and S. Sclaroff, “Translation and scale-invariant gesture recognition in complex scenes,” ACM PETRA '08
• 3D Maps, http://www.mapply.com/ , http://recce.at/
• IntelliVision, http://www.intelli-vision.com/products/intelligent-video-analytics/intelligent-video-motion-detector
• ActiServ: Activity Recognition Service for Mobile Phones, www.teco.kit.edu/~gordon/publications/ISWC10_berch.pdf
15
References