Résumé : In this talk, I will present the main ideas from this paper: Online Prediction in Sub-linear Space, B. Peng and F. Zhang, SODA 2023 (best student paper award). It gives new low-space algorithms for the regret minimisation problem in learning theory. Precious little is required in the way of prerequisites, except that you be intelligent enough to be interested in this sort of thing. The talk will start from the basics (the experts problems), and then cover the new ideas to entend classical learning algorithms to work, approximately, with sub-linear space.

 [arXiv]

Dernière modification : Tuesday 28 January 2025

Contact pour cette page : Cyril.Banderier at lipn.univ-paris13.fr