papers available on dimension reduction

[Kari Torkkola (a540aa)]

Two papers on dimension reduction are available:


1.
Kari Torkkola and William M. Campbell,
Mutual Information in Learning Feature Transformations

Abstract
We present feature transformations useful for exploratory data
analysis or for pattern recognition.  Transformations are learned from
example data sets by maximizing the mutual information between
transformed data and their class labels.  We make use of Renyi's
quadratic entropy, and we extend the work of Principe et al. to mutual
information between continuous multidimensional variables and
discrete-valued class labels.

The paper can be retrieved through page
http://members.home.net/torkkola/mmi.html
together with some illustrative examples.


2.
William M. Campbell, Kari Torkkola, and Sreeram V. Balakrishnan,
Dimension Reduction Techniques for Training Polynomial Networks

Abstract
We propose two novel methods for reducing dimension in training
polynomial networks. We consider the class of polynomial networks
whose output is the weighted sum of a basis of monomials. Our first
method for dimension reduction eliminates redundancy in the training
process.  Using an implicit matrix structure, we derive iterative
methods that converge quickly. A second method for dimension reduction
involves a novel application of random dimension reduction to
``feature space.'' The combination of these algorithms produces a
method for training polynomial networks on large data sets with
decreased computation over traditional methods and model complexity
reduction and control.

http://members.home.net/torkkola/sp_papers/campbell-icml2000.ps.gz or
http://members.home.net/torkkola/sp_papers/campbell-icml2000.pdf

Both papers will appear in the Proceedings of ICML 2000,
June 29 - July 2, Stanford, CA.


--
Kari Torkkola                      phone: +1-480-4134129
  
Motorola Labs, MD EL508            fax:   +1-480-4137281
2100 East Elliot Road              email: a540aa at email.mot.com
Tempe, AZ 85284                    http://members.home.net/torkkola