Publication:
Two Information-Theoretic Tools to Assess the Performance of Multi-class Classifiers

Thumbnail Image
Identifiers
URI: http://hdl.handle.net/10016/13118
ISSN: 0167-8655
DOI: 10.1016/j.patrec.2010.05.017
Files
PRL09_EMI_2col.pdf (340.48 KB)
Publication date
2010-09
Defense date
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Elsevier
Impact
Google Scholar
Export
Research Projects
Organizational Units
Journal Issue
Abstract
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.
Description
Keywords
Multi-class classifier, Confusion matrix, Contingency table, Performance measure, Performance measure, De Finetti diagram, Entropy triangle
Bibliographic citation
Pattern Recognition Letters, September 2010, Volume 31, Issue 12, pp. 1665-1671
Collections
DTSC - GPM - Artículos de Revistas