Semi-Supervised Learning of Mixture Models

Fabio Gagliardi Cozman, Ira Cohen, Marcelo Cesar Cirelo

Research output: Contribution to conferenceConference Paper

156 Scopus citations

Abstract

This paper analyzes the performance of semi-supervised learning of mixture models. We show that unlabeled data can lead to an increase in classification error even in situations where additional labeled data would decrease classification error. We present a mathematical analysis of this "degradation" phenomenon and show that it is due to the fact that bias may be adversely affected by unlabeled data. We discuss the impact of these theoretical results to practical situations.
Original languageAmerican English
Pages99-106
Number of pages8
StatePublished - 1 Dec 2003
Externally publishedYes
EventProceedings, Twentieth International Conference on Machine Learning -
Duration: 1 Dec 2003 → …

Conference

ConferenceProceedings, Twentieth International Conference on Machine Learning
Period1/12/03 → …

Fingerprint

Dive into the research topics of 'Semi-Supervised Learning of Mixture Models'. Together they form a unique fingerprint.

Cite this