Bayesian and Pairwise Local Similarity Discriminant Analysis

Home >> Publications >> CIP 2010 - Pairwise Local SDA Projects    Blog    Bio
Peter Sadowski, Luca Cazzanti and Maya R. Gupta

Abstract - We investigate three extensions to the generative similarity-based classifier called local similarity discriminant analysis (local SDA): a Bayesian approach to estimating the pmfs based on the assumption that similarities are multinomially distributed and on the Dirichlet prior distribution; a pairwise-similarity formulation of local SDA that accounts for all local pairwise similarities to estimate the pmfs; a combined Bayesian pairwise-similarity approach. We discuss how the proposed extensions afford more modeling flexibility than standard local SDA and less cumbersome model training than previously-published local SDA regularization strategies. Experiments with five benchmark similarity-based classification datasets show that the increased modeling flexibility and lighter computational burden of the proposed extensions are coupled with the good classification performance of the local SDA classification paradigm.

Datasets - AuralSonar | AmazonBinary | Patrol | Voting | Protein
Software - C++ source code for Bayesian local SDA using Dirichlet priors and pairwise local SDA, plus other classifiers we compaared to in the paper. Visit the downloads page for more software.

BibTex -
TITLE = {Bayesian and Pairwise Local Similarity Discriminant Analysis},
AUTHOR = {P. Sadowski and L. Cazzanti and M. R. Gupta},
BOOKTITLE = {Proc.Intl. Workshop on Cognitive Information Processing ({CIP})},
ADDRESS = {Isola d'Elba, Italy},
MONTH = {June},
YEAR = {2010}}