Two information-theoretic tools to assess the performance of multi-class classifiers

被引：16
|
作者：
Valverde-Albacete, Francisco J. [1 ]
Pelaez-Moreno, Carmen [1 ]
机构：
[1] Univ Carlos III Madrid, Dept Teoria Senal & Comunicac, Leganes 28911, Spain
来源：
PATTERN RECOGNITION LETTERS | 2010年 / 31卷 / 12期
关键词：
Multi-class classifier; Confusion matrix; Contingency table; Performance measure; de Finetti diagram; Entropy triangle; ROC CURVE; AREA;
D O I：
10.1016/j.patrec.2010.05.017
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
We develop two tools to analyze the behavior of multiple-class, or multi-class, classifiers by means of entropic measures on their confusion matrix or contingency table. First we obtain a balance equation on the entropies that captures interesting properties of the classifier. Second, by normalizing this balance equation we first obtain a 2-simplex in a three-dimensional entropy space and then the de Finetti entropy diagram or entropy triangle. We also give examples of the assessment of classifiers with these tools. (C) 2010 Elsevier B.V. All rights reserved.
引用
收藏
页码：1665 / 1671
页数：7
相关论文
共 50 条
  • [1] Information theoretic approach for performance evaluation of multi-class assignment systems
    Holt, Ryan S.
    Mastromarino, Peter A.
    Kao, Edward K.
    Hurley, Michael B.
    SIGNAL PROCESSING, SENSOR FUSION, AND TARGET RECOGNITION XIX, 2010, 7697
  • [2] Selection of classifiers using information-theoretic criteria
    Kang, HJ
    PATTERN RECOGNITION AND DATA MINING, PT 1, PROCEEDINGS, 2005, 3686 : 478 - 487
  • [3] On reoptimizing multi-class classifiers
    Bourke, Chris
    Deng, Kun
    Scott, Stephen D.
    Schapire, Robert E.
    Vinodchandran, N. V.
    MACHINE LEARNING, 2008, 71 (2-3) : 219 - 242
  • [4] On reoptimizing multi-class classifiers
    Chris Bourke
    Kun Deng
    Stephen D. Scott
    Robert E. Schapire
    N. V. Vinodchandran
    Machine Learning, 2008, 71 : 219 - 242
  • [5] INFORMATION-THEORETIC CRITERIA FOR THE DESIGN OF COMPRESSIVE SUBSPACE CLASSIFIERS
    Nokleby, Matthew
    Rodrigues, Miguel
    Calderbank, Robert
    2014 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2014,
  • [6] An Information-Theoretic Explanation for the Adversarial Fragility of AI Classifiers
    Xie, Hui
    Yi, Jirong
    Xu, Weiyu
    Mudumbai, Raghu
    2019 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY (ISIT), 2019, : 1977 - 1981
  • [7] Visualization and analysis of classifiers performance in multi-class medical data
    Diri, Banu
    Albayrak, Songul
    EXPERT SYSTEMS WITH APPLICATIONS, 2008, 34 (01) : 628 - 634
  • [8] Information-Theoretic Tools for Optical Communications Engineers
    Agrell, Erik
    Secondini, Marco
    2018 IEEE PHOTONICS CONFERENCE (IPC), 2018,
  • [9] Experimental comparisons of multi-class classifiers
    Institute of Intelligent Computing and Information Technology, Chengdu Normal University, No.99 East Haike Road, Wenjiang District, Chengdu, China
    不详
    Informatica, 1 (71-85):
  • [10] Experimental Comparisons of Multi-class Classifiers
    Li, Lin
    Li, Lin
    Wu, Yue
    Ye, Mao
    INFORMATICA-JOURNAL OF COMPUTING AND INFORMATICS, 2015, 39 (01): : 71 - 85
12345