Transcript Explaining Multivariate Time Series to Detect Early Problem Signs

Who am I and what am I
doing here?
Allan Tucker
A brief introduction to my research
www.brunel.ac.uk\~cssrajt
Outline of talk
My background
My current research and collaborations
A sample of results and publications
Plan of future research and funding
Conclusions
A bit of background
BSc Cognitive Science:
University of Sheffield, 1996
PhD Computer Science:
University of London, 2001
Post doctorate research:
Brunel University, 2001-2004
IDA group at Brunel
Headed by Professor Liu





Bioinformatics, Genomics, and Medical
Informatics
Data Mining and Intelligent Systems
Dynamic Systems and Signal Processing
Graphics, Images and Visualisation
Multivariate Time Series and Statistical
Analysis
Areas of interest
Bayesian networks
Automatic explanation of data
Multivariate time series
Classification
Optimisation
Collaborations
Moorfield’s eye hospital

Visual field understanding and classification
UCL, Department of virology

Gene expression data
Royal Holloway

Optimisation
Brunel University


Within IDA
Software engineering
One slide tutorial on Bayesian networks
Graph structure
Local probability distributions
Combine expert knowledge and data (but little
research on this)
P(B)
.001
Burglary
B
T
T
F
F
E
T
F
T
F
P(A)
.95
.94
.29
.001
Earthquake
P(E)
.002
Alarm
A P(M)
T .70
F .01
A
T
F
Mary Calls
John Calls
P(J)
.90
.05
Some results
Spatio-temporal models of visual fields

Artificial intelligence in medicine, 2004
Some results (continued)
Predicting Glaucoma
Some results (continued)
Explanation

Intelligent Data Analysis, 2002 & 2004
Some results (continued)
Combining expert knowledge and
data to identify relevant genes
Bioinformatics, under review
X04654
BC016182*
M85234
U32986
U82130
AK026463
323474
786846
Z14982*
U59309
D83785
47202
J03459
U87947
U77949
Z80783
BC009914
L21936
BC014433
U15173*
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
AK023995

Journal Publications
Tucker, A. Crampton, J. Swift, S. “RGFGA: An Efficient Representation and Crossover for
Grouping Genetic Algorithms” Evolutionary Computation, Provisionally Accepted.
Tucker, A. Vinciotti, V. Liu, X. Garway-Heath, D. “A Spatio-Temporal Bayesian Network
Classifier for Understanding Visual Field Deterioration”, Artificial Intelligence in
Medicine, Elsevier, In Press.
Swift, S. Tucker, A. Liu, X. Martin, N. Orengo, C. Kellam, P. “Consensus Clustering and
Functional Interpretation of Gene Expression Data”, Genome Biology, In Press.
Tucker, A. Vinciotti, V. Liu, X. “The Robust Selection of Predictive Genes Via a Simple
Classier”, Submitted to Bioinformatics.
Tucker, A and Liu, X “A Bayesian Network Approach to Explaining Time Series with
Changing Structure”, Intelligent Data Analysis – An International Journal, In Press.
Kellam, P. Liu, X. Martin, N. Orengo, C. Swift, S. Tucker, A. “A Framework for Modelling
Virus Gene Expression Data”, Intelligent Data Analysis, 2002.
Counsell, S. Liu, X. Mcfall, J. Swift, S. Tucker, A “Using Evolutionary Computation for
Clustering Email Data”, Intelligent Data Analysis, 2002.
Tucker, A. Liu, X. Ogden-Swift, A. “Evolutionary learning of dynamic probabilistic models
with large time lags”, International Journal of Intelligent Systems, 2001.
Swift, S. Tucker, A. Martin, N. Liu X. “Grouping Multivariate Time Series Variables:
Applications to Chemical process and Visual Field Data”, Knowledge Based Systems,
2001.
Tucker, A. Swift, S. Liu, X. “Grouping Multivariate Time Series via Correlation”, IEEE
Transactions on Systems, Man, and Cybernetics. Part B: Cybernetics, 2001.
Recent Conference Publications
Vinciotti, V. Tucker, A. Liu, X. Panteris, E. Kellam, P. “Identifying genes with high
confidence from small samples”, Workshop on Data Mining in Functional Genomics, at
the European Conference in Artificial Intelligence ECAI 2004.
Sheng, W. Tucker, A. Liu, X. “Clustering with Niching Genetic K-means Algorithm”,
GECCO 2004.
Tucker, A. Vinciotti, V. Liu, X. Garway-Heath, D. “Bayesian Networks to Classify Visual
Field Data”, The Association for Research in Vision and Ophthalmology Annual
Conference, ARVO 2004.
Tucker, A. Garway-Heath, D. Liu, X. “Bayesian Classification and Forecasting of Visual
Field Deterioration”, Proceedings of IDAMAP 2003.
Counsell, S., Liu, X., Najjar, R., Swift, S., Tucker, A., “Applying Intelligent Data Analysis to
Coupling Relationships in Object-oriented Software”, IDA 2003.
Tucker, A. Liu, X. “Learning Dynamic Bayesian Networks from Multivariate Time Series
with Changing Dependencies”, IDA 2003.
Tucker, A. Garway-Heath, D. Liu, X. “Spatial Operators for Evolving Dynamic Probabilistic
Networks from Spatio-Temporal Data”, GECCO 2003.
Counsell, S. Liu, X. McFall, J. Swift, S. and Tucker, A. “Optimising the Grouping of Email
Users to Serves Using Intelligent Data Analysis”, ICEIS 2001.
Kellam, P. Liu, X. Martin, N. Orengo, C. Swift, S. Tucker, A. “A Framework for Modelling
Short, High-Dimensional Multivariate Time Series: Preliminary Results in Virus Gene
Expression Data Analysis”, IDA 2001.
Tucker, A. Swift, S. Martin, N. Liu X. “Grouping Multivariate Time Series Variables:
Applications to Chemical process and Visual Field Data”, ES 2000.
Future directions
Continue existing research collaborations




Bioinformatics – HIV data, Gene identification
Software Engineering – Analysis of code
Optimisation – adaptive parameters, representations
Recently secured funding from Zeis Meditech in conjunction
with Moorfield’s to generate substantial data on visual fields
and retinal images
EPSRC first grant

Optimisation with adaptive parameters
BBSRC new investigation scheme

Combining databases (GO ENSEMBL) into coherent models
of the human genome
EPSRC advanced fellowship?
Summary
Record of working within Brunel over 4
years
Multiple projects and collaborations with
a number of institutions
Good publication record including
several “grade A” journals
Keen to build upon my research record
Thanks for listening
Any questions?
Some results
Clustering (MTS and Consensus)


IEEE System Man & Cybernetics, 2001
Genome Biology, 2004
Some results (continued)
Efficient representations for GAs


International Journal of Intelligent Systems, 2001
Evolutionary computation, provisionally accepted
-4000
Log Likelihood
-4500
-5000
EP
-5500
EP Seeded GA
(c=20%)
-6000
EP Seeded GA
(c=100%)
-6500
-7000
0
500
1000
1500
2000
Function Calls
2500
3000