Titre du document / Document title

Segmentation of multivariate mixed data via lossy coding and compression

Auteur(s) / Author(s)

DERKSEN Harm ⁽¹⁾ ; YI MA ⁽²⁾ ; WEI HONG ⁽³⁾ ; WRIGHT John ⁽²⁾ ;

Affiliation(s) du ou des auteurs / Author(s) Affiliation(s)

⁽¹⁾ Department of Mathematics, University of Michigan, ETATS-UNIS
⁽²⁾ Coordinated Science Laboratory, University of Illinois at Urbana Champaign, ETATS-UNIS
⁽³⁾ DSP Solutions Research and Development Center, Texas Instruments, ETATS-UNIS

Résumé / Abstract

In this paper, based on ideas from lossy data coding and compression, we present a simple but surprisingly effective technique for segmenting multivariate mixed data that are drawn from a mixture of Gaussian distributions or linear subspaces. The goal is to find the optimal segmentation that minimizes the overall coding length of the segmented data, subject to a given distortion. We show that deterministic segmentation minimizes an upper bound on the (asymptotically) optimal solution. The proposed algorithm does not require any prior knowledge of the number or dimension of the groups, nor does it involve any parameter estimation. Simulation results reveal intriguing phase-transition behaviors of the number of segments when changing the level of distortion or the amount of outliers. Finally, we demonstrate how this technique can be readily applied to segment real imagery and bioinformatic data.

Revue / Journal Title

Proceedings of SPIE, the International Society for Optical Engineering   ISSN 0277-786X   CODEN PSISDG 

Source / Source

Congrès
Visual communications and image processing 2007 :   ( 30 January-1 February, 2007, San Jose, California, USA )
Visual communications and image processing, San Jose CA , ETATS-UNIS (2007)
2007  , vol. 6508 (2), pp. 65080H.1-65080H.12[Note(s) : 2 vol., ] (12 ref.) ISBN 978-0-8194-6621-1 ;  Illustration : Illustration ;

Langue / Language

Anglais

Editeur / Publisher

Society of Photo-Optical Instrumentation Engineers, Bellingham, WA, ETATS-UNIS  (1981) (Revue)
SPIE, Bellingham Wa, ETATS-UNIS  (2007) (Monographie)

Mots-clés anglais / English Keywords

Outlier ; Simulation ; Parameter estimation ; Optimal solution ; Deterministic approach ; Subspace method ; Gaussian distribution ; Lossy medium ; Segmentation ; Algorithm ; Imagery ; Phase transitions ; Upper bound ; Coding ;

Mots-clés français / French Keywords

Observation aberrante ; Simulation ; Estimation paramètre ; Solution optimale ; Approche déterministe ; Méthode sous espace ; Loi normale ; Milieu dissipatif ; Segmentation ; Algorithme ; Imagerie ; Transition phase ; Borne supérieure ; Codage ;

Mots-clés espagnols / Spanish Keywords

Observación aberrante ; Simulación ; Estimación parámetro ; Solución óptima ; Enfoque determinista ; Método subespacio ; Curva Gauss ; Medio dispersor ; Segmentación ; Algoritmo ; Imaginería ; Transición fase ; Cota superior ; Codificación ;

Localisation / Location

INIST-CNRS, Cote INIST : 21760, 35400015359384.0160