VC theory of large margin multi-category classifiers

被引:0
|
作者
Guermeur, Yann [1 ]
机构
[1] LORIA, CNRS, F-54506 Vandoeuvre Les Nancy, France
关键词
multi-class discriminant analysis; large margin classifiers; uniform strong laws of large numbers; generalized VC dimensions; multi-class SVMs; structural risk minimization inductive principle; model selection;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
In the context of discriminant analysis, Vapnik's statistical learning theory has mainly been developed in three directions: the computation of dichotomies with binary-valued functions, the computation of dichotomies with real-valued functions, and the computation of polytomies with functions taking their values in finite sets, typically the set of categories itself. The case of classes of vector-valued functions used to compute polytomies has seldom been considered independently, which is unsatisfactory, for three main reasons. First, this case encompasses the other ones. Second, it cannot be treated appropriately through a naive extension of the results devoted to the computation of dichotomies. Third, most of the classification problems met in practice involve multiple categories. In this paper, a VC theory of large margin multi-category classifiers is introduced. Central in this theory are generalized VC dimensions called the gamma-Psi-dimensions. First, a uniform convergence bound on the risk of the classifiers of interest is derived. The capacity measure involved in this bound is a covering number. This covering number can be upper bounded in terms of the gamma-Psi-dimensions thanks to generalizations of Sauer's lemma, as is illustrated in the specific case of the scale-sensitive Natarajan dimension. A bound on this latter dimension is then computed for the class of functions on which multi-class SVMs are based. This makes it possible to apply the structural risk minimization inductive principle to those machines.
引用
收藏
页码:2551 / 2594
页数:44
相关论文
共 50 条
  • [21] A hybrid financial trading support system using multi-category classifiers and random forest
    Thakur, Manoj
    Kumar, Deepak
    APPLIED SOFT COMPUTING, 2018, 67 : 337 - 349
  • [22] Analysis of a multi-category classifier
    Anthony, Martin
    Ratsaby, Joel
    DISCRETE APPLIED MATHEMATICS, 2012, 160 (16-17) : 2329 - 2338
  • [23] A utility model for multi-category baskets
    Nadarajah, Saralees
    IMA JOURNAL OF MANAGEMENT MATHEMATICS, 2008, 19 (03) : 269 - 274
  • [24] Retrieval method for multi-category images
    Tanaka, H
    Sakano, H
    Ohtsuka, S
    PROCEEDINGS OF THE 17TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOL 2, 2004, : 965 - 968
  • [25] Models of Multi-Category Choice Behavior
    P. B. Seetharaman
    Siddhartha Chib
    Andrew Ainslie
    Peter Boatwright
    Tat Chan
    Sachin Gupta
    Nitin Mehta
    Vithala Rao
    Andrei Strijnev
    Marketing Letters, 2005, 16 : 239 - 254
  • [26] Multi-category demand and supermarket pricing
    Smith, Howard
    Thomassen, Oyvind
    INTERNATIONAL JOURNAL OF INDUSTRIAL ORGANIZATION, 2012, 30 (03) : 309 - 314
  • [27] Reliability of multi-category rating scales
    Parker, Richard I.
    Vannest, Kimberly J.
    Davis, John L.
    JOURNAL OF SCHOOL PSYCHOLOGY, 2013, 51 (02) : 217 - 229
  • [28] Models of multi-category choice behavior
    Seetharaman, PB
    Chib, S
    Ainslie, A
    Boatwright, P
    Chan, T
    Gupta, S
    Mehta, N
    Rao, V
    Strijnev, A
    MARKETING LETTERS, 2005, 16 (3-4) : 239 - 254
  • [29] A multi-category customer base analysis
    Park, Chang Hee
    Park, Young-Hoon
    Schweidel, David A.
    INTERNATIONAL JOURNAL OF RESEARCH IN MARKETING, 2014, 31 (03) : 266 - 279
  • [30] MCMOD: The Multi-Category Large-Scale Dataset for Maritime Object Detection
    Sun, Zihao
    Hu, Xiao
    Qi, Yining
    Huang, Yongfeng
    Li, Songbin
    CMC-COMPUTERS MATERIALS & CONTINUA, 2023, 75 (01): : 1657 - 1669