Skip to main navigation Skip to search Skip to main content

Coping with distribution change in the same domain using similarity-based instance weighting

  • Jeong Woo Son*
  • , Hyun Je Song
  • , Seong Bae Park
  • , Se Young Park
  • *Corresponding author for this work
  • Kyungpook National University

Research output: Contribution to conferenceConference paperpeer-review

Abstract

Lexicons are considered as the most crucial features in natural language processing (NLP), and thus often used in machine learning algorithms applied to NLP tasks. However, due to the diversity of lexical space, the machine learning algorithms with lexical features suffer from the difference between distributions of training and test data. In order to overcome the distribution change, this paper proposes support vector machines with example-wise weights. The training distribution coincides with the test distribution by weighting training examples according to their similarity to all test data. The experimental results on text chunking show that the distribution change between training and test data is actually recognized and the proposed method which considers this change in its training phase outperforms ordinary support vector machines.

Original languageEnglish
Title of host publicationAdvances in Machine Learning - First Asian Conference on Machine Learning, ACML 2009, Proceedings
Pages354-366
Number of pages13
DOIs
StatePublished - 2009
Event1st Asian Conference on Machine Learning, ACML 2009 - Nanjing, China
Duration: 2009.11.22009.11.4

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5828 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference1st Asian Conference on Machine Learning, ACML 2009
Country/TerritoryChina
CityNanjing
Period09.11.209.11.4

Fingerprint

Dive into the research topics of 'Coping with distribution change in the same domain using similarity-based instance weighting'. Together they form a unique fingerprint.

Cite this