This reprinting is corrected to software version 2.10

La diffusion de cette thèse se fait dans le respect des droits de son auteur, qui a signé le formulaire Autorisation de reproduire et de diffuser un travail de recherche de cycles supérieurs (SDU-522- Rév.01-2006). Cette autorisation stipule que «conformément à l'article 11 du Règlement no 8 des études de cycles supérieurs, [l'auteur] concède à l'Université du Québec à Montréal une licence non exclusive d'utilisation et de publication de la totalité ou d'une partie importante de [son] travail de recherche pour des fins pédagogiques et non commerciales. Plus précisément, [l'auteur] autorise l'Université du Québec à Montréal à reproduire, diffuser, prêter, distribuer ou vendre des copies de [son] travail de recherche à des fins non commerciales sur quelque support que ce soit, y compris l'Internet. Cette licence et cette autorisation n'entraînent pas une renonciation de [la] part [de l'auteur] à [ses] droits moraux ni à [ses] droits de propriété intellectuelle. Sauf entente contraire, [l'auteur] conserve la liberté de diffuser et de commercialiser ou non ce travail dont [il] possède un exemplaire. » REMERCIEMENTS Voici la petite paltie où l'on peut exprimer une goutte de sentiments et la subjectivité inhérente dans cet océan d'objectivité scientifique... Il n'est pas évident de rendre les

Maîtrise des caractéristiques des diodes électroluminescentes: un enjeu majeur en éclairage général

M. Yann VOITURON Professeur, Université Lyon1Signal honesty and condition-dependence in a multimodal sexual signaling system

This e-book is devoted to Global Optimization algorithms, which are methods for finding solutions of high quality for an incredible wide range of problems. We introduce the basic concepts of optimization and discuss features which make optimization problems difficult and thus, should be considered when trying to solve them. In this book, we focus on

Contribution bioinformatique à l'analyse du transcriptome humain Présentée et soutenue publiquement par Yann Loe-Mie

We describe a new incremental algorithm for training linear threshold functions: the Relaxed Online Maximum Margin Algorithm, or ROMMA. ROMMA can be viewed as an approximation to the algorithm that repeatedly chooses the hyperplane that classifies previously seen examples correctly with the maximum margin. It is known that such a maximum-margin hypothesis can be computed by minimizing the length of the weight vector subject to a number of linear constraints. ROMMA works by maintaining a relatively simple relaxation of these constraints that can be eciently updated. We prove a mistake bound for ROMMA that is the same as that proved for the perceptron algorithm. Our analysis implies that the more computationally intensive maximum-margin algorithm also satis es this mistake bound; this is the rst worst-case performance guarantee for this algorithm. We describe some experiments using ROMMA and a variant that updates its hypothesis more aggressively as batch algorithms to recognize handwr...

Existing statistical approaches to natural language problems are very coarse approximations to the true complexity of language processing. As such, no single technique will be best for all problem instances. Many researchers are examining ensemble methods that combine the output of successful, separately developed modules to create more accurate solutions. This paper examines three merging rules for combining probability distributions: the well known mixture rule, the logarithmic rule, and a novel product rule. These rules were applied with state-of-the-art results to two problems commonly used to assess human mastery of lexical semantics—synonym questions and analogy questions. All three merging rules result in ensembles that are more accurate than any of their component modules. The differences among the three rules are not statistically significant, but it is suggestive that the popular mixture rule is not the best rule for either of the two problems. 1

Abstract not found

Emily Oster, and Fabrizio Zilibotti for helpful comments. We also thank for their comments seminar participants at the University of Chicago, CREI, Harvard, Zurich and to the NBER Cultural Economics workshop. Yann Algan thanks the ERC-Starting Grant for financial support. The views expressed herein