Système de classification hybride interprétable par construction automatique de systèmes d'inférence floue

Abstract : In pattern recognition, it is often difficult to perform classifiers that are at the same time accurate, generic and that model knowledge in a comprehensible way. Nevertheless, ``transparent'' classifiers allow experts to maintain systems and to operate several optimizations (recognition rates, complexity, ...). To satisfy these objectives, we propose a new hybrid classifier that takes advantage of a double modeling based on two different kind of knowledge: intrinsic and discriminant. The knowledge is extracted automatically and clarified in a robust way by fuzzy sets. These ones are used to generate fuzzy inference systems that allow aggregation and fusion of the knowledge to make decision in a robust and comprehensive way. Experimentations reported here also validate adaptability and accuracy of the classifier
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01191664
Contributor : Nicolas Ragot <>
Submitted on : Thursday, September 3, 2015 - 2:34:35 PM
Last modification on : Tuesday, July 2, 2019 - 4:02:03 PM

Links full text

Identifiers

Citation

Nicolas Ragot, Eric Anquetil. Système de classification hybride interprétable par construction automatique de systèmes d'inférence floue. Revue des Sciences et Technologies de l'Information - Série TSI : Technique et Science Informatiques, Lavoisier, 2003, n° spécial : Fusion numérique/symbolique, 22 (7-8), pp.853-878. ⟨10.3166/tsi.22.853-878⟩. ⟨hal-01191664⟩

Share

Metrics

Record views

402