938 citations
- The Alan Turing InstituteGB178 papers
- Queen Mary University of LondonGB41 papers
- University College LondonGB36 papers
- University of CambridgeGB35 papers
- Imperial College LondonGB32 papers
- University of OxfordGB31 papers
- University of WarwickGB31 papers
- University of ManchesterGB25 papers
- University of EdinburghGB20 papers
- University of SouthamptonGB14 papers
- University of LondonGB11 papers
- City, University of LondonGB10 papers
19 papers · 1 filter
Non-Vacuous Generalization Bounds: Can Rescaling Invariances Help?
Damien Rouchouse, Antoine Gonon, Rémi Gribonval +1
A central challenge in understanding generalization is to obtain non-vacuous guarantees that go beyond worst-case complexity over data or weight space. Among existing approaches, P…
Learning via Surrogate PAC-Bayes
Antoine Picard-Weibel, Roman Moscoviz, Benjamin Guedj
PAC-Bayes learning is a comprehensive setting for (i) studying the generalisation ability of learning algorithms and (ii) deriving new learning algorithms by optimising a generalis…
A note on regularised NTK dynamics with an application to PAC-Bayesian training
Eugenio Clerico, Benjamin Guedj
We establish explicit dynamics for neural networks whose training objective has a regularising term that constrains the parameters to remain close to their initial value. This keep…
Invariant Causal Imitation Learning for Generalizable Policies
Ioana Bica, Daniel Jarrett, Mihaela van der Schaar
Consider learning an imitation policy on the basis of demonstrated behavior from multiple environments, with an eye towards deployment in an unseen environment. Since the observabl…
Time-series Generation by Contrastive Imitation
Daniel Jarrett, Ioana Bica, Mihaela van der Schaar
Consider learning a generative model for time-series data. The sequential setting poses a unique challenge: Not only should the generator capture the conditional dynamics of (stepw…
Wasserstein PAC-Bayes Learning: Exploiting Optimisation Guarantees to Explain Generalisation
Maxime Haddouche, Benjamin Guedj
PAC-Bayes learning is an established framework to both assess the generalisation ability of learning algorithms, and design new learning algorithm by exploiting generalisation boun…