Title: Online Bayesian Learning of Sparse Text Classifiers
Speakers: Suhrid Balakrishnan and David Madigan, Rutgers University
Date: Monday, March 21, 2005 4:15pm
Location: DIMACS Center, CoRE Bldg, Room 433, Rutgers University, Busch Campus, Piscataway, NJ
Abstract:
Classification of text data is a challenging task, complicated by the very high dimensionality of (and size of existing labeled) data sets. Classifiers favoring sparse solutions (such as the SVM, RVM) have proven to be the most competitive methods for this problem thus far. This talk will present ongoing work outlining the development of an online regression based classifier that favors sparse solutions. Related work includes assumed density filtering, Bayesian parameter estimation by variational approximation and the online Bayesian algorithm presented in Opper '99.