Semi-supervised clustering with pairwise constraints

A discriminative approach

Zhengdong Lu, Todd K. Leen

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

We consider the semi-supervised clustering problem where we know (with varying degree of certainty) that some sample pairs are (or are not) in the same class. Unlike previous efforts in adapting clustering algorithms to incorporate those pairwise relations, our work is based on a discriminative model. We generalize the standard Gaussian process classifier (GPC) to express our classification preference. To use the samples not involved in pairwise relations, we employ the graph kernels (covariance matrix) based on the entire data set. Experiments on a variety of data sets show that our algorithm significantly outperforms several state-of-the-art methods.

Original languageEnglish (US)
Pages (from-to)299-306
Number of pages8
JournalJournal of Machine Learning Research
Volume2
StatePublished - 2007

Fingerprint

Semi-supervised Clustering
Covariance matrix
Clustering algorithms
Pairwise
Classifiers
Gaussian Process
Clustering Algorithm
Express
Experiments
Classifier
Entire
kernel
Generalise
Graph in graph theory
Experiment
Model
Class
Standards

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software
  • Control and Systems Engineering
  • Statistics and Probability

Cite this

Semi-supervised clustering with pairwise constraints : A discriminative approach. / Lu, Zhengdong; Leen, Todd K.

In: Journal of Machine Learning Research, Vol. 2, 2007, p. 299-306.

Research output: Contribution to journalArticle

@article{a39f61cce9d242beaf6d188b89b24bbe,
title = "Semi-supervised clustering with pairwise constraints: A discriminative approach",
abstract = "We consider the semi-supervised clustering problem where we know (with varying degree of certainty) that some sample pairs are (or are not) in the same class. Unlike previous efforts in adapting clustering algorithms to incorporate those pairwise relations, our work is based on a discriminative model. We generalize the standard Gaussian process classifier (GPC) to express our classification preference. To use the samples not involved in pairwise relations, we employ the graph kernels (covariance matrix) based on the entire data set. Experiments on a variety of data sets show that our algorithm significantly outperforms several state-of-the-art methods.",
author = "Zhengdong Lu and Leen, {Todd K.}",
year = "2007",
language = "English (US)",
volume = "2",
pages = "299--306",
journal = "Journal of Machine Learning Research",
issn = "1532-4435",
publisher = "Microtome Publishing",

}

TY - JOUR

T1 - Semi-supervised clustering with pairwise constraints

T2 - A discriminative approach

AU - Lu, Zhengdong

AU - Leen, Todd K.

PY - 2007

Y1 - 2007

N2 - We consider the semi-supervised clustering problem where we know (with varying degree of certainty) that some sample pairs are (or are not) in the same class. Unlike previous efforts in adapting clustering algorithms to incorporate those pairwise relations, our work is based on a discriminative model. We generalize the standard Gaussian process classifier (GPC) to express our classification preference. To use the samples not involved in pairwise relations, we employ the graph kernels (covariance matrix) based on the entire data set. Experiments on a variety of data sets show that our algorithm significantly outperforms several state-of-the-art methods.

AB - We consider the semi-supervised clustering problem where we know (with varying degree of certainty) that some sample pairs are (or are not) in the same class. Unlike previous efforts in adapting clustering algorithms to incorporate those pairwise relations, our work is based on a discriminative model. We generalize the standard Gaussian process classifier (GPC) to express our classification preference. To use the samples not involved in pairwise relations, we employ the graph kernels (covariance matrix) based on the entire data set. Experiments on a variety of data sets show that our algorithm significantly outperforms several state-of-the-art methods.

UR - http://www.scopus.com/inward/record.url?scp=84862279781&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84862279781&partnerID=8YFLogxK

M3 - Article

VL - 2

SP - 299

EP - 306

JO - Journal of Machine Learning Research

JF - Journal of Machine Learning Research

SN - 1532-4435

ER -