A feature selection method based on Choquet Integral and Typicality Analysis

Cyril Mazaud (1) , Jan Rendek (2) , Vincent Bombardier (1) , Laurent Wendling (2)

1 CRAN - Centre de Recherche en Automatique de Nancy
2 QGAR - Querying Graphics through Analysis and Recognition

In this paper, we present an iterative feature selection method based on feature typicality and interactivity analysis. The aim of such a method is to enhance model interpretability by selecting the best significant features among a list extracted from images. The inference mechanism uses a fuzzy linguistic rule-based system. In the presented application, we apply this method to wood defect classification. Nowadays, feature selection is expertise-driven and most of the time, expert uses features by habits which not always represent the best ones to use. The proposed method aims to replace expert selection by automatically choosing the most adapted features to the recognition problem.

Domaines: Traitement du signal et de l'image [eess.SP]; Automatique / Robotique; Traitement du signal et de l'image [eess.SP]

Cyril Mazaud, Jan Rendek, Vincent Bombardier, Laurent Wendling. A feature selection method based on Choquet Integral and Typicality Analysis. 16th International Conference on Fuzzy Systems - FUZZIEEE'07, Jul 2007, Londres, United Kingdom. pp.1703-1708, ⟨10.1109/FUZZY.2007.4295623⟩. ⟨hal-00137983⟩