Post on 22-Mar-2017
transcript
STAND AND DELIVER:HOW TO GIVE A GREAT RESEARCH TALK
Dr. Julie GreensmithIntelligent Modelling and Analysis Research GroupSchool of Computer Science University of Nottingham
LETS ELIMINATE BORING TALKS
We all have to give research talks in order to communicate our scientific research
Just as critical as being able to write a good paper
Generate impact for our research
Encourage others to be interested in our ideas
Gain esteem within our community and in industry
To inspire the next generation of researchers
COMMUNICATION IS ESSENTIAL
Great ideas are nothing, literally worthless, if you dont communicate them to others
Used to crystallise ideas and communicate to your community
Gain feedback on research and ideas
Build relationships with others in your field and the wider community
THE PURPOSE OF THIS TUTORIAL
To describe how to deliver a great research talk
To engage others in your research
Discuss the purpose of giving research talks
Ways to achieve engaging delivery
Not to give a tutorial on how to design slides or on powerpoint!
INSPIRATION
Simon Peyton Jones (SPJ) of Microsoft Research, Cambridge, UK
Based on a 1993 Research Paper [1]
Also has written some excellent guides to writing research papers and also grant proposals
Plus he does that thing called “functional programming”
http://research.microsoft.com/en-us/um/people/simonpj/papers/giving-a-talk/giving-a-talk.htm
WHY GIVE RESEARCH TALKS?
We really should focus on journal publications with high impact factors (as far as the REF is concerned!)
So, why do we spend years of our lives preparing, writing and presenting conference papers?
The presentation of either your poster or talk is the culmination of the work
Not just so we can travel on expenses…
WHAT YOUR TALK IS FOR
Your paper is the ‘meat’ of the work
Your talk is the advertisement for the ‘meat’
SPJ says:
“Never Confuse The Two”
YOUR GOAL
To entice the listener to want to go away and read your paper
to follow on your work
to cite your work in future
Its not a showcase to show everything that you have ever learned about your thesis topic
Not necessary to go into excruciating detail
YOUR PERFECT AUDIENCE
Have read all of your previous papers and all of the papers that you reference in all your papers
Thorough understanding of all previously published artificial immune systems and algorithms and a perfect working knowledge of the history of immunology and contemporary immunology.
And will be able to understand equations and pseudocode in a single glance
Are fresh, raring to go, eager to hear you speak and love your powerpoint template
YOUR ACTUAL AUDIENCE
Have never heard of you
Are not aware of all the literature ever published in your field
Have just had lunch and are drowsy
Have jet-lag
Are desperate to check their email, facebook, twitter
TALK STRUCTURE
1. Motivation for the research - 20%
2. The key idea of the research - 80%
3. “There is no number 3” - [SPJ]
Not:
a rehash of your paper translated into powerpoint slides
its a totally different communication medium
CRITICAL PERIOD
How long do you think you have before the audience decides to listen to your talk (or not)?
This first period is critical, it is when the listener will decide if they want to invest their attention in what you are saying
You must excite them in the first couple of minutes
laptops start opening, Facebook statuses get updated
Similar to the introduction of a research paper
IN THE FIRST TWO MINUTES
Engage the audience to let them know your motivation:
why they should invest energy in listening to you
what your problem is and why it is interesting
This means don’t start your talk with a lengthy description of your field. It wont interest and excite
Instead, start with the problem first
EXAMPLE - PROBLEM FIRST
Artificial Immune Systems including the Negative Selection Algorithm have been applied to a variety of problems within computer security
However these systems can be prone to poor scaling and the generation of false positive alarms
In this work we attempt to construct a new type of artificial immune system which can scale and has much lower rates of false positives
We will capture novel immunology to do this
PRESENTING THE KEY IDEA
“If you remember nothing else I say today, remember this…”
Get it absolutely clear in your head before you start writing your slides what is your key idea
if you are not clear about this, your audience wont be either
Be specific - its your job to figure this out, not the audience
Be explicit - say “this is the key idea of the paper”
Be precise - make sure the rest of the talk is structured around communicating this
DEPTH OF TALK
“Narrow and deep beats superficial and broad”
Avoid presenting a condensed summary of the paper
Use a problem centred approach to focus your mind and theirs
Examples are the best way to do this
Remember, your talk is an “advert for the meat”
EXAMPLES ARE YOUR WEAPON
“When time is short, omit the general case and not the example”
Use them to motivate the work and to convey the basic intuition of the idea
Help you to illustrate ‘The Idea’ in action
To show extreme and interesting cases
To highlight shortcomings and limitations
Relate it to something the audience already understands
USING EXAMPLES
Want to describe the signal processing aspects of a new algorithm
Can go into the mathematical detail and biological metaphor for as long as you like - but it doesn't help
Before presenting the algorithm, describe a specific scenario in which you will frame the work
Use the example to illustrate the idea and to make it relatable
DECENT EXAMPLE
If we capture data representing an internal intrusion in a LAN we can derive the following signals
Signal 1:
Number of network packets sent per second as an indication of the activity of the host machine
Represents the “danger signal”
Signal 2:
The rate of ‘burstiness’ of sending of network packets as an indication of the type of network connection
Represents the “safe signal”
BAD EXAMPLE
Danger signals:
measure of an attribute which significantly increases in response to abnormal behaviour
a moderate degree of confidence of abnormality with increased level of this signal, though at a low signal strength can represent normal behaviour.
Safe signals:
a high rate of change equals a low safe signal level and vice versa.
a confident indicator of normal behaviour in a predictable manner or a measure of steady- behaviour
measure of an attribute which increases signal concentration due to the lack of change in strength
IN INTERDISCIPLINARY WORK
Both research groups perform a lot of interdisciplinary research
Use a trivial example to relate the concepts from one domain into another
requires a good idea of to which audience you are presenting
“How could I demonstrate this idea to my mum?”
APOPTOSIS VS NECROSIS (BIO)
The innate immune system can tell the difference between a tissue fullof necrotic product and a tissue containing apoptotic cells
Necrosis results in the release of internal cell contents which degrade and become danger signals (Olpan et al 2010).
apoptotic context yields semi-mature cells
necrotic context yields mature cells
It is not the presence of antigen but the detection of damage which isthe initial activator of the human immune system
APOPTOSIS VS NECROSIS (CS)
A computationally interesting mechanism
A practical demonstration of how this works
Gives the audience a much better grounding of what is going on
THE POINTLESS OVERVIEW
A short history of Artificial Immune Systems
1st Generation Algorithms
2nd Generation Algorithms
Interdisciplinary development of immune-inspired algorithms
Systems Not Algorithms
Emerging research in artificial immune ensembles
LEAVE IT OUT!
It conveys nearly zero information about your talk to the listener
Eats into the precious two minutes of attention - its boring!!
Either is too superficial to provide any information or you have to explain too much and half the talk is delivered from just this slide
Use signpost slides throughout the talk instead to give inherent structure to the presentation
I hate the ‘slide countdown’ in the corner, it makes me desperate for the talk to end and makes me count down the slides until I can have a coffee!!
THE PAPER EQUIVALENT
How many of you actually read these paragraphs in papers?
“In this paper The DCA is applied to the detection of a port scan, which forms a convenient small-scale computer security problem. Section 2 contains relevant background information regarding the problem of port scans and current scanning detection techniques. Section 3 presents the biological inspiration of the DCA, a summary of relevant developments in immunology, and rudimentary DC biology. This is followed by Sections 4 and 5, describing the abstraction process, a formalised description of the DCA and its implementation as an anomaly detector. This is followed by experimentation with its application as a port scan detector. Section 6 includes a sensitivity analysis of a selection of parameters. The paper concludes with a discussion of the results of the port scan investigation and suggestions for future work.” - Greensmith et al (2010).
17 SLIDES ON RELATED WORK
You don’t need to show off how many papers you have read in a 15 minute research talk
Maximum one slide showing the evolution of the work
Include relevant references in the main body slides
You must absolutely know the related work and be able to answer questions on it
Be careful not to destructively criticise other’s work (they might be in the room!!)
TALKING-TOO-TECHNICAL
We put it in as it represents hours of hard work and blood
If given two minutes there is no way I can understand reams of equations
Just showing equations for the sake of looking clever is not
Present only the parts that are relevant to the talk itself
interested parties will always refer to the paper
NEVER APOLOGISE
“I didn’t have time to prepare this talk properly”
“My computer broke down, so I don’t have the results I expected”
“I don’t have time to tell you about this”
“I don’t feel qualified to address this audience”
“I couldn’t find the graph I was looking for”
“I’m sorry my English is not good”
PREPARATION
Prepare your slides as close to the time of the talk as you dare
They must be totally fresh in your mind
Practice your talk on anyone who will listen/video yourself
Keep your slides as simple as you can so you are not distracted
Avoid reading off cue cards/random bits of paper
Never, ever, ever, simply read off your slides
ENTHUSIASM
If you dont look excited by the subject, why would anyone else?
It makes people more receptive to your ideas if you yourself look convinced of the matter
Energy to move, point, breathe also helps overcome nervousness
It wakes people up - the main goal!
MANAGING ANXIETY
Even the most seasoned professionals can get ‘stage fright’
Jelly legs, wobbly voice, can’t breathe, inability to operate brain
Go to the loo before your session starts
Have a bottle of water handy as your mouth will dry up
Wear something that will not show up massive underarm sweat patches
ANXIETY IS NORMAL
Take on as much oxygen as you can, it will calm your body down
Script your first few sentences precisely, so you dont need your brain until you have got going
Move about and be animated, it seems to burn off the adrenaline
If your hands are shaking, dont use a lazer pointer
You are not alone or weak - we all feel like this!
BE SEEN AND HEARD
Dont hide behind your laptop, it makes you look even more scared and point at the screen and not your laptop
Try and speak to the back of the room
Identify your “nodders” and make eye contact with them
Dont be put off if someone looks unhappy
they probably just checked their email and had a grant proposal rejected, or have indigestion from some dodgy conference food
Powerpoint Annoyances
Use animations very sparingly
Avoid noises of any kind
Check your videos will work
Make sure you have the right dongle
ANSWERING QUESTIONS
Many people, sometimes prestigious, fall apart at this bit
Questions are a golden golden golden opportunity to connect with your audience
Specifically encourage questions during your talk: pause briefly now and then, ask for questions
Ask someone to rephrase their question as a stalling tactic
TIMING
Never overrun - it looks really bad
At a conference where time is tight, it is important not to overrun - practicing for timing really helps
Audiences will stop listening when time runs out
Never ask “can I carry on” - its awkward and hard for the chair of the session to say no
Unlike this tutorial where I will do what I want.
SEE WHAT YOU HAVE LEARNED
Write down all of the problems you can see in this presentation
https://www.youtube.com/watch?v=nSGqp4-bZQY
RAISING THE STANDARD
The standard of conference presentations is sometimes so low, that it is a great opportunity to make a good impression just by using one or two of the tips presented today
“The standard is so low you don't have to be
outstanding to stand out” - [SPJ]
PROACTIVE PRACTICE
Been sent on numerous lecturer ‘boot camp’ activities including how to make yourself a better presenter
Most of the advice was not useful or practical
Best activity was to prepare and present a 5 minute talk which was videotaped - highlighted all sorts of weird things I was unaware of!
Also the same for making research based videos for computerphile
THIS AFTERNOON’S ACTIVITY
Using a smartphone camera, in pairs or threes, make a 2 minute video about some interesting aspect of your research
Good practice at presenting but also really need some new promotional IMA material!
Editing does not have to be massively slick, but would like to showcase some ideas at the end of today’s session
We can help do some of the editing - now film the footage :)
I WOULDN'T MAKE YOU DO ANYTHING I WOULDN'T DO MYSELF
https://www.youtube.com/watch?v=CyZ38n9flrE