Étude des mécanismes d'adaptation et de rejet pour l'optimisation de classifieurs : Application à la reconnaissance de l'écriture manuscrite en-ligne

Harold Mouchère 1
1 IMADOC - Interprétation et Reconnaissance d’Images et de Documents
UR1 - Université de Rennes 1, INSA Rennes - Institut National des Sciences Appliquées - Rennes, CNRS - Centre National de la Recherche Scientifique : UMR6074
Abstract : The aim of these work is to increase the classifier accuracy using two approaches : rejection and adaptation. The rejection allows to decide if the classifier answer is relevant or not. The diversity of the applications requiring this concept makes us to distinguish two main natures of reject with distinct goals: the confusion reject and the distance reject. We present a unified formalism AMTL to define them using reliability functions and reject thresholds. We also present an automatic on-line adaptation mechanism (ADAPT) to the writer's handwriting style for the recognition of isolated handwritten characters. The classifier is based on a fuzzy inference system (FIS). This FIS is composed of fuzzy prototypes which represent the intrinsic properties of the classes and it uses numeric conclusions. The proposed adaptation mechanism affects both the conclusions of the rules and the fuzzy prototypes of the premises by re-centering and re-shaping them. To increase the adaptation speed we propose to generate new data from previous samples inputed by the user. To validate these approaches, we propose several experimentation from simulations to real condition tests.
Document type :
Theses
Complete list of metadatas

Cited literature [122 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00379228
Contributor : Harold Mouchère <>
Submitted on : Tuesday, April 28, 2009 - 9:47:36 AM
Last modification on : Friday, November 16, 2018 - 1:28:16 AM
Long-term archiving on : Thursday, June 10, 2010 - 10:03:41 PM

Identifiers

  • HAL Id : tel-00379228, version 1

Citation

Harold Mouchère. Étude des mécanismes d'adaptation et de rejet pour l'optimisation de classifieurs : Application à la reconnaissance de l'écriture manuscrite en-ligne. Interface homme-machine [cs.HC]. INSA de Rennes, 2007. Français. ⟨tel-00379228⟩

Share

Metrics

Record views

309

Files downloads

726