Least-Squares Conditional Density Estimation

• SUGIYAMA Masashi

  Tokyo Institute of Technology PRESTO, Japan Science and Technology Agency

• TAKEUCHI Ichiro

  Nagoya Institute of Technology

• SUZUKI Taiji

  The University of Tokyo

• KANAMORI Takafumi

  Nagoya University

• HACHIYA Hirotaka

  Tokyo Institute of Technology

• OKANOHARA Daisuke

  The University of Tokyo

Abstract

Estimating the conditional mean of an input-output relation is the goal of regression. However, regression analysis is not sufficiently informative if the conditional distribution has multi-modality, is highly asymmetric, or contains heteroscedastic noise. In such scenarios, estimating the conditional distribution itself would be more useful. In this paper, we propose a novel method of conditional density estimation that is suitable for multi-dimensional continuous variables. The basic idea of the proposed method is to express the conditional density in terms of the density ratio and the ratio is directly estimated without going through density estimation. Experiments using benchmark and robot transition datasets illustrate the usefulness of the proposed approach.

Journal

• IEICE Transactions on Information and Systems

  IEICE Transactions on Information and Systems E93-D (3), 583-594, 2010

  The Institute of Electronics, Information and Communication Engineers

Keywords

• conditional density estimation
• multimodality
• heteroscedastic noise
• direct density ratio estimation
• transition estimation

Details

• CRID: 1390001204379401984
• NII Article ID: 10026814428
• NII Book ID: AA10826272
• DOI: 10.1587/transinf.e93.d.583
• ISSN: 17451361
• ISSN: 09168532
• Web Site: http://www.jstage.jst.go.jp/article/transinf/E93.D/3/E93.D_3_583/_pdf
• Text Lang: en
• Data Source:
  • JaLC
  • Crossref
  • CiNii Articles
• Abstract License Flag: Disallowed