The National Program on
Complex Data Structures

 
       

Workshop on Data Mining Methodology and Applications
October 28-30, 2004
at The Fields Institute, Toronto
Supported by

Tentative Schedule

Thursday October 28, 2004

8:30-9:00 REGISTRATION
COFFEE AND CONTINENTAL BREAKFAST
9:00- 9:20 Opening Remarks
9:20-10:20 Helmut Kröger
Learning in neural networks with small-world architecture.
10:20-10:45 COFFEE
10:45-12:15

Rare target problems

Stan Young
Linking and pattern matching in multiple large data two-way tables
Mu Zhu
An Adaptive Radial Basis Function Network Model for Statistical Detection"
Grigoris Karakoulas

ROC-based Learning for Imbalanced Class Problems

12:15-1:45 LUNCH
1:45-3:15

Unsupervised methods I

Steven Wang
Clustering Categorical Data Based on Distance Vectors
Russel Steele,
Algebraic Geometry and Model Selection for Naive Bayes Networks
Xianping Liu
Generation 5 Hybrid Clustering System and its Application

3:15-3:45 COFFEE/TEA BREAK
3:45-4:45

Feature extraction

Roberto Aldave and Simon Gluzman
Prediction of Real Variables with Non-Polynomial Approximants
Wenxue Huang
Dependence Degree and Feature Selection for Categorical Data

4:45-5:15 Daily discussant: William Welch, UBC
5:15-7:00 Reception hosted by

Friday October 29, 2004

8:30-9:00 COFFEE AND CONTINENTAL BREAKFAST
9:00-10:00 Jerome Friedman
Importance Sampling: An Alternative View of Ensemble Learning*
*Joint work with Bogdan Popescu
10:00-10:30 COFFEE
10:30-12:00

SAMSI data mining theme year speakers

David Banks
Scalability of Models in Data Mining
Adele Cutler
Random Forests: Proximity, Variable Importance, and Visualization*
*Joint work with Leo Breiman
Merlise Clyde
Bayesian Perspectives on Combining Models

12:00-1:30 LUNCH
1:30-3:30

Supervised methods I

Alex Depoutovitch
The use of grid computing to speed up prediction
Reuben Zamar
Robust Methods and Data Mining
Godfried Toussaint
Proximity Graph Methods for Data Mining
Alex Zolotovitski

Automated Trade area analysis. Case study of G5 MWM software application

3:30-4:00 COFFEE/TEA BREAK
4:00-5:00 Ji Zhu and Saharon Rosset
Is regularization: efficient and effective
Piecewise linear SVM paths
5:00-5:30 Daily discussant: Hugh Chipman, Acadia University
5:30 -7:00 Reception hosted by the National Program on Complex Data Structures (cash bar)

Saturday October 30, 2004

8:30-9:00 COFFEE AND CONTINENTAL BREAKFAST
9:00-10:00 Yoshua Bengio
Statistical Learning from High Dimensional and Complex Data: Not a Lost Cause
10:00-10:30 Coffee
10:30-12:00

Mining industrial process data

Joaquin Ordieres Meré
Data-Mining for industrial processes
Theodora Kourti,
Data Mining in Industry for Process and Product Improvement

12:00-1:30 LUNCH
1:30-3:30

Panel Discussion

Tracey Jarosz, Loyalty group
Jerome Friedman, Stanford University
Theodora Kourti, McMaster University
Rick Makos, Teradata
Ivan Miletic, Dofasco Inc.
Milorad Krneta, Generation 5
Stan Young, National Institute of Statistical Sciences, North Carolina

3:30-4:00

COFFEE/TEA BREAK

4:00-5:00 Djamel Zighed
Constructing induction graphs