A generalization of principal component analysis to the exponential family

A generalization of principal component analysis to the exponential family (2001)

by Michael Collins , Sanjoy Dasgupta , Robert E. Schapire

Venue:	Advances in Neural Information Processing Systems
Citations:	86 - 0 self

Datum	Value	Source
TITLE	A generalization of principal component analysis to the exponential family	INFERENCE
AUTHOR NAME	Michael Collins	SVM HeaderParse 0.2
AUTHOR AFFIL	AT&T Labs Research	SVM HeaderParse 0.2
AUTHOR ADDR	180 Park Avenue, Florham Park, NJ 07932	SVM HeaderParse 0.2
AUTHOR NAME	Sanjoy Dasgupta	SVM HeaderParse 0.2
AUTHOR AFFIL	AT&T Labs Research	SVM HeaderParse 0.2
AUTHOR ADDR	180 Park Avenue, Florham Park, NJ 07932	SVM HeaderParse 0.2
AUTHOR NAME	Robert E. Schapire	SVM HeaderParse 0.2
AUTHOR AFFIL	AT&T Labs Research	SVM HeaderParse 0.2
AUTHOR ADDR	180 Park Avenue, Florham Park, NJ 07932	SVM HeaderParse 0.2
ABSTRACT	Principal component analysis (PCA) is a commonly applied technique for dimensionality reduction. PCA implicitly minimizes a squared loss function, which may be inappropriate for data that is not real-valued, such as binary-valued data. This paper draws on ideas from the Exponential family, Generalized linear models, and Bregman distances, to give a generalization of PCA to loss functions that we argue are better suited to other data types. We describe algorithms for minimizing the loss functions, and give examples on simulated data. 1	SVM HeaderParse 0.2
YEAR	2001	INFERENCE
VENUE	Advances in Neural Information Processing Systems	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
VOLUME	14	INFERENCE
CITATIONS	10 found	ParsCit 1.0