Liste de publications

Conférences Internationales

[1] G. Gavin, S. Di Palma et D.A. Zighed. Generalisation properties of convex linear combinations of classifiers belonging to finite VC-dimension family. Fourth International Conference on Computer Science and Informatics, octobre 1998.

[2] G. Gavin et A. Elisseeff. Confidence bounds for the generalization performances of linear combination of functions., Ninth International Conference on Artificial Neural Networks (ICANN),University of Edinburgh, UK: 7-10, september 1999.

[3] G. Gavin, D. Puzenat et D.A. Zighed. How to use linear combinations of functions to get classifiers with low decision making error. In proceedings of « International Conference on Artificial Intelligence » (ICAI’99), Las Vegas, juin 1999.

[4] G. Gavin. Confidence bounds for one hidden-layer perceptrons. To appear in proccedings of “International Conference on Artificial Intelligence and Soft Computing”, Cancun, may 2001.

[5] G. Gavin, O. Teytaud. Lowers bounds on the sample complexity. Submitted to” International Joint Conference on Neural Networks”, Wahington, september 2001.

[6] G. Gavin. Confidence bounds for multi-layers perceptrons. Submitted to “Internationnal Conference on Artificial Neural Networks”, Vienne, august 2001.

Conférences nationales avec comité de lecture

[7] G. Gavin. Etude de l’apprentissage Probablement Approximativement Correct : Application aux méthodes d’apprentissage. These de Doctorat. 2001.

[8] G. Gavin, R. Rakotomalala et D.A. Zighed. Heuristique de pondération en vue de l’amélioration des performances du bagging. Société Francophone de Classification, 1997.

[9] G. Gavin. Apprentissage supervisé et agrégation de prédicteurs. Diplome d’études Approfondies en Informatique. 1997.

Rapports de recherche