Towards self-exploring discriminating features

Ying Wu, Thomas S. Huang

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Many visual learning tasks are usually confronted by some common difficulties. One of them is the lack of supervised information, due to the fact that labeling could be tedious, expensive or even impossible. Such scenario makes it challenging to learn object concepts from images. This problem could be alleviated by taking a hybrid of labeled and unlabeled training data for learning. Since the unlabeled data characterize the joint probability across different features, they could be used to boost weak classifiers by exploring discriminating features in a self-supervised fashion. Discriminant-EM (D-EM) attacks such problems by integrating discriminant analysis with the EM framework. Both linear and nonlinear methods are investigated in this paper. Based on kernel multiple discriminant analysis (KMDA), the nonlinear D-EM provides better ability to simplify the probabilistic structures of data distributions in a discrimination space. We also propose a novel data-sampling scheme for efficient learning of kernel discriminants. Our experimental results showthat D-EM outperforms a variety of supervised and semi-supervised learning algorithms for many visual learning tasks, such as content-based image retrieval and invariant object recognition.

Original languageEnglish (US)
Title of host publicationMachine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings
EditorsPetra Perner
PublisherSpringer Verlag
Number of pages15
ISBN (Print)3540423591, 9783540423591
StatePublished - 2001
Event2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001 - Leipzig, Germany
Duration: Jul 25 2001Jul 27 2001

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2123 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Other2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Towards self-exploring discriminating features'. Together they form a unique fingerprint.

Cite this