Visualizing class structure in data using mutual information

Author

Torkkola, Kari

Author_Institution

Motorola Inc., Tempe, AZ, USA

Volume

1

fYear

2000

fDate

2000

Firstpage

376

Abstract

We study linear dimension reducing transforms using maximum mutual information between transformed data and class labels as the criterion to learn the transforms. Renyi quadratic entropy provides a differentiable and computationally feasible criterion on which gradient ascent algorithms can be based without the limitations of methods using only second order statistics, such as PCA or LDA. Application to class structure visualization in exploratory data analysis is presented

Keywords

data analysis; data reduction; data visualisation; entropy; information theory; neural nets; pattern recognition; Renyi quadratic entropy; class labels; class structure; class structure visualization; exploratory data analysis; gradient ascent algorithms; linear dimension reducing transforms; mutual information; transformed data; transforms; Covariance matrix; Data analysis; Data visualization; Eigenvalues and eigenfunctions; Entropy; Independent component analysis; Linear discriminant analysis; Mutual information; Principal component analysis; Statistics;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks for Signal Processing X, 2000. Proceedings of the 2000 IEEE Signal Processing Society Workshop

Conference_Location

Sydney, NSW

ISSN

1089-3555

Print_ISBN

0-7803-6278-0

Type

conf

DOI

10.1109/NNSP.2000.889429

Filename

889429