OREGON STATE UNIVERSITY

You are here

Learning non-redundant codebooks for classifying complex objects

TitleLearning non-redundant codebooks for classifying complex objects
Publication TypeConference Paper
Year of Publication2009
AuthorsZhang, W., A. Surve, X. Z. Fern, and T. G. Dietterich
Tertiary AuthorsDanyluk, A., L. Bottou, and M. Littman
Conference NameProceedings of the 26th Annual International Conference on Machine Learning - ICML '09
Pagination1 - 8
Date Published06/2009
PublisherACM Press
Conference LocationMontreal, Quebec, Canada
ISBN Number9781605585161
Abstract

Codebook-based representations are widely employed in the classification of complex objects such as images and documents. Most previous codebook-based methods construct a single codebook via clustering that maps a bag of low-level features into a fixed-length histogram that describes the distribution of these features. This paper describes a simple yet effective framework for learning multiple non-redundant codebooks that produces surprisingly good results. In this framework, each codebook is learned in sequence to extract discriminative information that was not captured by preceding codebooks and their corresponding classifiers. We apply this framework to two application domains: visual object categorization and document classification. Experiments on large classification tasks show substantial improvements in performance compared to a single codebook or codebooks learned in a bagging style.

DOI10.1145/1553374.1553533