Year: 2025
Type: preprint
Abstract: PAC generalization bounds on the risk, when expressed in terms of the expected loss, are often insufficient to capture imbalances between subgroups in the data. To overcome this limitation, we introduce a new family of risk measures, called constrained f-entropic risk measures, which enable finer control over distributional shifts and subgroup imbalances via f-divergences, and include the Conditional Value at Risk (CVaR), a well-known risk measure. We derive both classical and disintegrated PAC-Bayesian generalization bounds for this family of risks, providing the first disintegratedPAC-Bayesian guarantees beyond standard risks. Building on this theory, we design a self-bounding algorithm that minimizes our bounds directly, yielding models with guarantees at the subgroup level. Finally, we empirically demonstrate the usefulness of our approach.
Source: arXiv (Cornell University)
Institutions: Centre National de la Recherche Scientifique, Laboratoire Hubert Curien, Centre Inria de Lyon, MALICE: Apprentissage automatique avec intégration des connaissances en ingénierie de surface : théorie et algorithmes, Université Jean Monnet, Université Jean Monnet, Université Lumière Lyon 2, Université de Lyon, Entrepôts, Représentation et Ingénierie des Connaissances, Institut national de recherche en sciences et technologies du numérique, Institut de Recherche en Informatique et Systèmes Aléatoires, MALT: Apprentissage Automatique avec Contraintes Temporelles, Université de Rennes
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: Agence Nationale de la Recherche
Award: Fair Multimodal Learning
OpenAlex ID: W4415249790
Cites:
Cited by:
Related to: