Combining Discriminant Models with new Multi-Class SVMs
Résumé
The idea of combining models instead of simply selecting the ``best'' one, in order to improve performance, is well known in statistics and has a long theoretical background. However, making full use of theoretical results is ordinarily subject to the satisfaction of strong hypotheses (weak correlation among the errors, availability of large training sets, possibility to rerun the training procedure an arbitrary number of times, etc.). In contrast, the practitioner who has to make a decision is frequently faced with the difficult problem of combining a given set of pretrained classifiers, with highly correlated errors, using only a small training sample. Overfitting is then the main risk, which cannot be overcome but with a strict complexity control of the combiner selected. This suggests that SVMs, which implement the SRM inductive principle, should be well suited for these difficult situations. Investigating this idea, we introduce a new family of multi-class SVMs and assess them as ensemble methods on a real-world problem. This task, protein secondary structure prediction, is an open problem in biocomputing for which model combination appears to be an issue of central importance. Experimental evidence highlights the gain in quality resulting from combining some of the most widely used prediction methods with our SVMs rather than with the ensemble methods traditionally used in the field. The gain is increased when the outputs of the combiners are post-processed with a simple DP algorithm.
Mots clés
modèles hiérarchiques de traitement de séquences
hierarchical sequence processing systems
combinaison de modèles
classifier fusion
statistical learning theory
generalization performance
support vector machines
protein secondary structure prediction
théorie de l'apprentissage statistique
performances en généralisation
machines à vecteurs support
prédiction de la structure secondaire des protéines