eprintid: 10138174 rev_number: 20 eprint_status: archive userid: 608 dir: disk0/10/13/81/74 datestamp: 2021-11-15 12:02:20 lastmod: 2021-11-15 12:02:20 status_changed: 2021-11-15 12:02:20 type: proceedings_section metadata_visibility: show creators_name: Zantedeschi, V creators_name: Viallard, P creators_name: Morvant, E creators_name: Emonet, R creators_name: Habrard, A creators_name: Germain, P creators_name: Guedj, B title: Learning Stochastic Majority Votes by Minimizing a PAC-Bayes Generalization Bound ispublished: pub divisions: UCL divisions: B04 divisions: C05 divisions: F48 note: This version is the version of record. For information on re-use, please refer to the publisher's terms and conditions. abstract: We investigate a stochastic counterpart of majority votes over finite ensembles of classifiers, and study its generalization properties. While our approach holds for arbitrary distributions, we instantiate it with Dirichlet distributions: this allows for a closed-form and differentiable expression for the expected risk, which then turns the generalization bound into a tractable training objective.The resulting stochastic majority vote learning algorithm achieves state-of-the-art accuracy and benefits from (non-vacuous) tight generalization bounds, in a series of numerical experiments when compared to competing algorithms which also minimize PAC-Bayes objectives -- both with uninformed (data-independent) and informed (data-dependent) priors. date: 2021 date_type: published publisher: Advances in Neural Information Processing Systems (NeurIPS 2021) official_url: https://proceedings.neurips.cc/paper/2021/hash/0415740eaa4d9decbc8da001d3fd805f-Abstract.html oa_status: green full_text_type: pub language: eng primo: open primo_central: open_green verified: verified_manual elements_id: 1872870 lyricists_name: Guedj, Benjamin lyricists_id: BGUED94 actors_name: Guedj, Benjamin actors_id: BGUED94 actors_role: owner full_text_status: public publication: CoRR event_title: 35th Conference on Neural Information Processing Systems (NeurIPS 2021) book_title: Proceedings of the 35th Conference on Neural Information Processing Systems (NeurIPS 2021) citation: Zantedeschi, V; Viallard, P; Morvant, E; Emonet, R; Habrard, A; Germain, P; Guedj, B; (2021) Learning Stochastic Majority Votes by Minimizing a PAC-Bayes Generalization Bound. In: Proceedings of the 35th Conference on Neural Information Processing Systems (NeurIPS 2021). Advances in Neural Information Processing Systems (NeurIPS 2021) Green open access document_url: https://discovery.ucl.ac.uk/id/eprint/10138174/7/Guedj_Learning%20Stochastic%20Majority%20Votes%20by%20Minimizing%20a%20PAC-Bayes%20Generalization%20Bound_VoR.pdf