Bagging de caractéristiques pour l'authentification d'auteur

The authorship attribution literature demonstrates the difficulty to design classifiers that outperform simple strategies such as linear classifiers operating on bag of features representation of documents. To overcome this difficulty we propose to use Bagging techniques that rely on learning classifiers on different random subsets of features, then to combine their decision by making them vote. MOTS-CLÉS : Authentification d’auteur, Bagging de caractéristiques