A simple algorithm for predicting nearly as well as the best pruning labeled with the best prediction values of a decision tree

Eiji Takimoto, Ken'ichi Hirai, Akira Maruoka

研究成果: Chapter in Book/Report/Conference proceedingConference contribution

3 被引用数 (Scopus)

抄録

Given an unpruned decision tree, Helmbold and Schapire gave an on-line prediction algorithm whose performance will not be much worse than the predictions made by the best pruning of the given decision tree. In this paper, based on the observation that finding the best pruning can be efficiently solved by a dynamic programming in the “batch” setting where all the data to be predicted are given in advance, a new on-line prediction algorithm is constructed. Although it is shown that its performance is slightly weaker than Helmbold and Sehapire’s algorithm with respect to the loss bound, it is so simple and general that it could be applied to many on-line optimization problems solved by dynamic programming. We also explore algorithms that are competitive not only with the best pruning but also with the best prediction values. In this setting, a greatly simplified algorithm is given, and it is shown that the algorithm can easily be generalized to the case where, instead of using decision trees, data are classified in some arbitrarily fixed manner.

本文言語英語
ホスト出版物のタイトルAlgorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings
編集者Ming Li, Akira Maruoka
出版社Springer Verlag
ページ385-400
ページ数16
ISBN(印刷版)3540635777, 9783540635772
DOI
出版ステータス出版済み - 1 1 1997
外部発表はい
イベント8th International Workshop on Algorithmic Learning Theory, ALT 1997 - Sendai, 日本
継続期間: 10 6 199710 8 1997

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
1316
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

その他

その他8th International Workshop on Algorithmic Learning Theory, ALT 1997
Country日本
CitySendai
Period10/6/9710/8/97

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

フィンガープリント 「A simple algorithm for predicting nearly as well as the best pruning labeled with the best prediction values of a decision tree」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル