Spatial Operators for Evolving Dynamic Bayesian Networks from Spatio-Temporal Data Allan Tucker Xiaohui Liu David Garway-Heath Moorfields Eye Hospital.

Slides:



Advertisements
Similar presentations
Introduction to Graphical Models Brookes Vision Lab Reading Group.
Advertisements

Ranking Outliers Using Symmetric Neighborhood Relationship Wen Jin, Anthony K.H. Tung, Jiawei Han, and Wei Wang Advances in Knowledge Discovery and Data.
Aggregating local image descriptors into compact codes
Danzhou Liu Ee-Peng Lim Wee-Keong Ng
Cost-effective Outbreak Detection in Networks Jure Leskovec, Andreas Krause, Carlos Guestrin, Christos Faloutsos, Jeanne VanBriesen, Natalie Glance.
Dynamic Bayesian Networks (DBNs)
UNIVERSITY OF JYVÄSKYLÄ Building NeuroSearch – Intelligent Evolutionary Search Algorithm For Peer-to-Peer Environment Master’s Thesis by Joni Töyrylä
The purpose of this study is to use statistical and classification models to classify, detect and understand progression in visual fields (VFs) We intend.
Introduction of Probabilistic Reasoning and Bayesian Networks
Three-Dimensional Teaching Study on the College Statistics Education Tengzhong Rong, Qiongsun Liu Chongqing university, China
Bayesian networks and how they can help us to explore fish species interaction in the Northern gulf of St Lawrence Dr Allan Tucker Centre for Intelligent.
1 A Framework for Modelling Short, High-Dimensional Multivariate Time Series: Preliminary Results in Virus Gene Expression Data Analysis Paul Kellam 1,
Integrating Bayesian Networks and Simpson’s Paradox in Data Mining Alex Freitas University of Kent Ken McGarry University of Sunderland.
1 Structure of search space, complexity of stochastic combinatorial optimization algorithms and application to biological motifs discovery Robin Gras INRIA.
Estimation of Distribution Algorithms Ata Kaban School of Computer Science The University of Birmingham.
1 Grouping Multivariate Time Series Variables: Applications to Chemical Process and Visual Field Data Allan Tucker- Birkbeck College Stephen Swift- Brunel.
Knowledge Engineering a Bayesian Network for an Ecological Risk Assessment (KEBN-ERA) Owen Woodberry Supervisors: Ann Nicholson Kevin Korb Carmel Pollino.
Are you being blinded by statistics? The structure of the retinal nerve fibre layer.
Spatio-Temporal Databases
Object-based Image Representation Dr. B.S. Manjunath Sitaram Bhagavathy Shawn Newsam Baris Sumengen Vision Research Lab University of California, Santa.
Genetic algorithms for neural networks An introduction.
Spatio-Temporal Databases. Outline Spatial Databases Temporal Databases Spatio-temporal Databases Multimedia Databases …..
Extending Evolutionary Programming to the Learning of Dynamic Bayesian Networks Allan Tucker Xiaohui Liu Birkbeck College University of London.
UNIVERSITY OF JYVÄSKYLÄ Topology Management in Unstructured P2P Networks Using Neural Networks Presentation for IEEE Congress on Evolutionary Computing.
Improvements in the Spatial and Temporal representation of the Model Owen Woodberry Bachelor of Computer Science, Honours.
Learning Dynamic Bayesian Networks with Changing Dependencies Allan Tucker Xiaohui Liu IDA 2003.
EA* A Hybrid Approach Robbie Hanson. What is it?  The A* algorithm, using an EA for the heuristic.  An efficient way of partitioning the search space.
Who am I and what am I doing here? Allan Tucker A brief introduction to my research
The Automatic Explanation of Multivariate Time Series (MTS) Allan Tucker.
Bridging the Gap between Applications and Tools: Modeling Multivariate Time Series X Liu, S Swift & A Tucker Department of Computer Science Birkbeck College.
Explaining Multivariate Time Series to Detect Early Problem Signs Architectures and Efficient Learning Algorithms for Dynamic Bayesian Networks Allan Tucker,
Learning Bayesian Networks
D Nagesh Kumar, IIScOptimization Methods: M1L4 1 Introduction and Basic Concepts Classical and Advanced Techniques for Optimization.
Computer vision: models, learning and inference Chapter 10 Graphical Models.
Bayesian Classification and Forecasting of Visual Field Deterioration Allan Tucker, Xiaohui Liu; Brunel University David Garway-Heath; Moorfield’s Eye.
Results Comparison with existing approaches on benchmark datasets Evaluation on a uveal melanoma datasetEvaluation on the two-spiral dataset Evaluation.
Mobile Filtering for Error-Bounded Data Collection in Sensor Networks Dan Wang Hong Kong Polytechnic Univ. Jianliang Xu ∗ Hong Kong Baptist Univ. Jiangchuan.
Nogood Recording for Static and Dynamic Constraint Satisfaction Problems Thomas Schiex, Gerard Verfaillie C.E.R.T.-O.N.E.R.A.(France)
Mobile Robot ApplicationsMobile Robot Applications Textbook: –T. Bräunl Embedded Robotics, Springer 2003 Recommended Reading: 1. J. Jones, A. Flynn: Mobile.
Breeding Decision Trees Using Evolutionary Techniques Papagelis Athanasios - Kalles Dimitrios Computer Technology Institute & AHEAD RM.
WAES 3308 Numerical Methods for AI
Gridding Daily Climate Variables for use in ENSEMBLES Malcolm Haylock, Climatic Research Unit Nynke Hofstra, Mark New, Phil Jones.
Extending Spatial Hot Spot Detection Techniques to Temporal Dimensions Sungsoon Hwang Department of Geography State University of New York at Buffalo DMGIS.
KNOWLEDGE BASED TECHNIQUES INTRODUCTION many geographical problems are ill-structured an ill-structured problem "lacks a solution algorithm.
Genetic Algorithms Siddhartha K. Shakya School of Computing. The Robert Gordon University Aberdeen, UK
CASE STUDY: NEW NAMING CONVENTION IN SHAREPOINT David Schlachter.
Crash Cube: An application of Map Cube to Hotspot Discovery in Vehicle Crash Data Mark Dietz, Jesse Vig CSCI 8715 Spatial Databases University of Minnesota.
Workshop on Stock Assessment Methods 7-11 November IATTC, La Jolla, CA, USA.
Evolutionary Programming
Making Time: Pseudo Time-Series for the Temporal Analysis of Cross-Section Data Emma Peeling, Allan Tucker Centre for Intelligent Data Analysis Brunel.
Interactive Evolution in Automated Knowledge Discovery Tomáš Řehořek March 2011.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 A self-organizing map for adaptive processing of structured.
Rate of Visual Field Progression in Eyes With Optic Disc Hemorrhages in the Ocular Hypertension Treatment Study De Moraes CG, Demirel S, Gardiner SK, et.
04/21/2005 CS673 1 Being Bayesian About Network Structure A Bayesian Approach to Structure Discovery in Bayesian Networks Nir Friedman and Daphne Koller.
1 Structure Learning (The Good), The Bad, The Ugly Inference Graphical Models – Carlos Guestrin Carnegie Mellon University October 13 th, 2008 Readings:
Function BIRN The ability to find a subject who may have participated in multiple experiments and had multiple assessments done is a critical component.
An Evolutionary Algorithm for Neural Network Learning using Direct Encoding Paul Batchis Department of Computer Science Rutgers University.
Estimation of Distribution Algorithm and Genetic Programming Structure Complexity Lab,Seoul National University KIM KANGIL.
Privacy Vulnerability of Published Anonymous Mobility Traces Chris Y. T. Ma, David K. Y. Yau, Nung Kwan Yip (Purdue University) Nageswara S. V. Rao (Oak.
Michal Irani Dept of Computer Science and Applied Math Weizmann Institute of Science Rehovot, ISRAEL Spatio-Temporal Analysis and Manipulation of Visual.
Lecture 3: Uninformed Search
DM-Group Meeting Liangzhe Chen, Nov
Model Averaging with Discrete Bayesian Network Classifiers
Probability and Time: Markov Models
Noémi Gaskó, Rodica Ioana Lung, Mihai Alexandru Suciu
Incorporating Constraint Checking Costs in Constraint Satisfaction Problem Suryakant Sansare.
Cost-effective Outbreak Detection in Networks
GATree Genetically Evolved Decision Trees
Probability and Time: Markov Models
Coevolutionary Automated Software Correction
Presentation transcript:

Spatial Operators for Evolving Dynamic Bayesian Networks from Spatio-Temporal Data Allan Tucker Xiaohui Liu David Garway-Heath Moorfields Eye Hospital NHS Trust

Contents of Talk Introduction to BNs, DBNs, and SDBNs Visual Field Data Representation and Spatial Operators The Experiments Results (Inc. Demo of the Operators) Conclusions

BNs, DBNs and SDBNs

Visual Field Data Collected From an Extensive Study Investigating OHT VF Tests carried out approximately every month 54 Points on the VF including two on the Blind Spot 95 Patients (1809 measurements in all)

Visual Field Data

The Datasets Visual Field Data 54 Variables, 95 Patients, 1809 Time Points Synthetic Data 64 DBN Variables Representing 8x8 Grid Parents: 1 st Order Cartesian Neighbours with Time Lag of 1 Each Node has Gaussian CPT

Representation and Operators Population Represents the Solution Individual Represents Point in Space and its Dependencies Efficient Use of Calls to Fitness Spatial, Non-Spatial and Temporal Operators Applied to Individuals

Representation {{a x,a y,l}, {a x,a y,l}, {a x,a y,l}} {{a x,a y,l}, {a x,a y,l}} {{a x,a y,l}, {a x,a y,l}, {a x,a y,l}}

Spatial Operators

The Experiments Spatial Operators Only Non-Spatial Operators Only Both Sets of Operators Investigate Learning Curves (Log-Lik) and Operator Success Rate Compare to Strawman Greedy Search Investigate SD, and Expert Knowledge

Results – Synthetic Data Spatial Operators Only Perform the Best Non-Spatial and K2 are the Worst Non-Spatial Appears to Eventually Discover a ‘Good’ Structure

Results – Synthetic Data Most Successful Operator by far is SpatAdd Take, and SpatMut are also Good SpatCross Looks Bad (Few Successes’) But Accounts for Biggest Fitness Improvements

Results – Visual Field Data This Time All- Operators Performs Best Closely Followed by Spatial Only But Given Time Non Spatial Catch Up K2 Performs Very Poorly

Results – Visual Field Data Again SpatAdd, Take, and SpatMut are Best SpatCross Looks Better But Still Least Successes Again Accounts for Biggest Fitness Improvements

Results

Spatial Operator Demo 1

Spatial Operator Demo 2

Spatial Operator Demo 3

Spatial Operator Demo 4

Spatial Operator Demo 5

Conclusions Developed Evolutionary Operators Specifically Designed for Spatial Data Efficient Representation Perform Competitively Compared to Standard Operators on Synthetic and Real World Data Generates VF SDBNs Consistent with Experts

Future Work Explore Other Spatial Datasets e.g. Rainfall Investigate Other Methods Developed for Spatial NN Function – EDAs Extend the VF Model to Include Both Eyes and Clinical Information

Any Questions?