PR Links

Feature Selection

Multidimensional Scaling:

	Webpage by Stephen P. Bourgatti, with good description and example.
	Matlab example using the statistics toolbox

Wrappers (Kohavi and John)

	Basic paper describing wrapper approach
	More detailed paper with more thorough description (44 pages)

Filters

	Information theory based approach to feature selection (RELIEF)
	Correlation-based feature selection, comparison to wrappers

Classifier Comparison and Tools

Bootstrap and applications; nice article from IEEE Signal Processing Magazine, very accessible and readable, by Zoubir and Boashash.

Decision Trees Tutorials and Software

Overview of Decision Trees, focuses on ID3: nice worked out examples used in class

Tutorial on building ID3 and C4.5 decision trees, Temple University

Quinlan's university web page, has papers and code for c4.5

Quinlan's company Rulequest.com with new improved tree building routines (See5).

Notes on Bagging and Boosting Classifiers

	Bagging, Breiman, technical report providing theoretical development of bagging.
	Boosting, Recent tutorial article by Schapire on Boosting, its history and variations.
	Applying Bagging and Boosting to C4.5, written by Quinlan in 1996.

Support vector machines background

	Burges, C.J.C, "A Tutorial on Support Vector Machines for Pattern Recognition," Data Mining and Knowledge Discovery, Volume 2, number 2, pages 121-167, 1998.
	Hsu, C.W, Chang, C.C., Lin, C.J., "A Practical Guide to Support Vector Classification,"
	Kernal Machines site
	Scholkopf NIPS 2001 Tutorial
	Scholkopf Kernal Machines review paper.
	In Class SVM demos (use libsvm)

Tools

	GGobi - Statistical visualization package for exploring data.
	R: open source statistical software package
	Matlab - The statistical, neural network, and fuzzy logic toolbox will be very useful for processing data for homework and projects.
	Classification Toolbox-written to support the textbook. This toolbox started as a course assignment in Dr. Ron Meir’s graduate course, Pattern Recognition at Technion – Israel Institute of Technology. The foundation for the toolbox, as well as most of the basic algorithms, were coded by Elad Yom-Tov and Hilit Serby. A year later, Igor Makienko and Victor Yosef coded the Voted Perceptron algorithms. More about the toolbox here.
	Bootstrap Toolbox by Zoubir and Iskander.

Data Sets and Repositories

STATLIB. Datasets from the Statistics Department at CMU.

UCI Machine Learning Data Repository

If you publish material based on databases obtained from the UCI repository, then, in your acknowledgments, please note the assistance you received by using this repository. This will help others to obtain the same data sets and replicate your experiments.

Blake, C.L. & Merz, C.J. (1998). UCI Repository of machine learning databases [http://www.ics.uci.edu/~mlearn/MLRepository.html]. Irvine, CA: University of California, Department of Information and Computer Science.