Change-point estimation using new Minimum Message Length approximations

Leigh J Fitzgibbon, David L Dowe, Lloyd Allison

    Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

    6 Citations (Scopus)


    This paper investigates the coding of change-points in the information-theoretic Minimum Message Length (MML) framework. Change-point coding regions affect model selection and parameter estimation in problems such as time series segmentation and decision trees. The Minimum Message Length (MML) and Minimum Description Length (MDL78) approaches to change-point problems have been shown to perform well by several authors. In this paper we compare some published MML and MDL78 methods and introduce some new MML approximations called ‘MMLDc’ and ‘MMLDF’. These new approximations are empirically compared with Strict MML (SMML), Fairly Strict MML (FSMML), MML68, the Minimum Expected Kullback-Leibler Distance (MEKLD) loss function and MDL78 on a tractable binomial change-point problem.
    Original languageEnglish
    Title of host publicationPRICAI 2002: Trends in Artificial Intelligence
    Subtitle of host publication7th Pacific Rim International Conference on Artificial Intelligence Tokyo, Japan, August 18-22, 2002 Proceedings
    EditorsMitsuru Ishizuka, Abdul Sattar
    Place of PublicationBerlin Germany
    Number of pages11
    ISBN (Print)3540440380
    Publication statusPublished - 2002
    EventPacific Rim International Conference on Artificial Intelligence 2002 - Tokyo, Japan
    Duration: 18 Aug 200222 Aug 2002
    Conference number: 7th (Proceedings)

    Publication series

    NameLecture Notes in Computer Science
    ISSN (Print)0302-9743


    ConferencePacific Rim International Conference on Artificial Intelligence 2002
    Abbreviated titlePRICAI 2002
    Internet address

    Cite this