Transcript Les Ball

Dr Les Ball
Intelligent Systems
Division Of Computing, School of Computing, Engineering and Applied Mathematics, University of Abertay Dundee.
Intelligence & CyberSecurity
Social Network Analysis
My research interests are centred on Intelligent Systems and their
ability to enhance decision-making processes by quantitative
analyses of intelligence data. Intelligence data relate predominantly
to issues of national security but also concern commercial
applications and health informatics. In security, applications are
focused on preventative measures of what we term the ‘pre-attack’
phase (see diagram below). This involves the acquisition of biometric
data and intelligence gathering from open sources. Understanding
the ‘mood’ of someone also plays a part in assessing threats to
security, while also finding application in customer satisfaction during
purchasing transactions and in matters of health diagnosis where a
patient may exhibit changes in their everyday behaviour. My work
specifically addresses the questions ‘who are you?’, ‘who do you
know?’, ‘what do you think?’ and ‘how do you feel?’.
Keyboard Dynamics & Stress Signatures
Did you know that your keyboard typing rhythm (dwell and flight times) is
your digital equivalent to your handwritten signature? Well, we can also
analyse and determine whether you are under stress when you are typing
from the changes in your rhythm.
http://www.inmagine.com/phins
104/phins104098-photo
Dowman M, Szymkowiak A, Coull N, Hapca S, Ball L.
Gesture Recognition
The post-event Social Network Analysis model of the 9/11 attacks showing hijacker pilots
(squared), densest section (left dashed box) incorporates 3 of the pilots and all of Mohamed
Atta’s associates (circles), the chained structure of UA#175 associates (triangles), Hani
Hanjour and associates (rhombi) are distinctly separated from the other hijacker pilots (right
dashed box). Source: http://orgnet.com/tnet.html
Your facial expressions can tell us a lot about the type of emotion you are
feeling by the detection of Facial Action Units. Experimental data below
show that most people purse their lips when experience trouble using selfservice checkouts i.e. a feeling of frustration.
What Do You Think?
60
Related to lip press
50
The vision of the School’s Security Research Group identifies three phases on a temporal basis
Who Are You?
Fingerprints have a long history in terms of identification. However,
with improvements in technology biometric software is now capable of
acquiring and analysing high-resolution data for identification and
verification purposes. Moreover, multimodal biometrics systems
research is now coming to the fore whereby more accurate
predictions are made by combining, say, face recognition with
fingerprint data as below. These systems are very important in high
security domains such as at airports and other border controls.
architecture for a computer system to perform automatic extraction from
social media to analyse political opinion.
Sentiment Analysis/Opinion Mining
40
Occurrences
Textual data are being deposited exponentially at social networking sites
such as Twitter and Facebook. This presents an opportunity to analyse
these data to answer questions such as “what do people think about our
new product?” or “what is current opinion on the issue of Scottish
independence?”. Opinion mining and sentiment analysis are two
synonymous techniques that use computational linguistics to solve
these problems and offer rapid results that would otherwise be
extremely time consuming to organise. The diagram below shows a typical
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Facial Action Unit
Martin C, Ball L, Archibald J.
eHealth Indicators
By sensing an outpatient’s data from everyday use such as movement
within the house, keyboard usage or eye movements, we might detect a
deviation or divergence from your normal pattern, which may in itself be
an indicator of a change in health condition e.g. the onset of Alzheimer’s.
Sensors
Recorder(s)
Processing
Gesture, action, etc
Biometrics
Outcomes
Ball L, Bradley D, Szymkowiak A, Brownsell S.
References
How Do You Feel?
Ball L, Bradley D, Brownsell S: 2011 Emotive Computing may have a Role in
Telecare, Journal of Telemedicine and Telecare, 17(6): 279-280.
We know that computers are great at number crunching and that they
are also getting better at detecting who you are. Have you however ever
wondered though if your computer might be able to understand how you
feel? Affecting Computing is a discipline addressing this very question.
We show what is called human ‘affect’ when we are under stress or in
an emotional state. So, wouldn’t it be good if a computer could detect
your mood by analysing your actions or expressions? The following
three diagrams illustrate how keyboard dynamics, facial expression
recognition and behaviour change could be used within the contexts of
security, retail shopping and eHealth.
Ball L, Bradley D, Szymkowiak A, Brownsell S: 2011 Linking Recorded
Data with Emotive and Adaptive Computing in eHealth, IEEE HISB 2011,
26-29 July San Jose, 198-204.
www.hitachi.com
Who Do You Know?
Social network analysis has been around for a long time within the
social sciences. It is now, however, being applied to digital data from
open sources. Its significance is that it can bring to light the emergence
of a new criminal or terrorist cell in terms of its communications
structure. If the key actors and substructures can be identified then
intelligence can strategically target their elimination and disrupt or
weaken the network. The following diagram was constructed by Valdis
Krebs from open source information after the 9/11 attacks.
About the Author
Dr Les Ball is a lecturer within the
School of Computing, Engineering
and Applied Mathematics. He
teaches principally how computer
algorithms can be used to enhance
decision-making processes in the
fields of Intelligence and Security
Informatics and eHealth.
Poster template by ResearchPosters.co.za
Dowman M, Szymkowiak A, Coull N, Hapca S, Ball L: 2011 A Biometric
System for Identifying Individuals and their emotional states from
keystroke metrics, in prep for submission to PNAS.
Martin C, Ball L, Archibald J: 2011 Emotion recognition in Self-Service
Interactions, Third International Conference on Intelligent Human Computer
Interaction, 29-31 August, Prague.
Contact Me
For further information
about the author use your
smart phone to capture the
QR code or go to
http://www.abertay.ac.uk
Dr Les Ball.
School of Computing, Engineering and Applied Mathematics,
University of Abertay Dundee,
Bell Street,
Dundee, DD1 1HG.
T : 01382 308251
E : [email protected]