GT Transport optimal - EDP - Machine learning
Mean-field Langevin dynamics and Neural Networks
17
June 2024
-
17
June 2024
June 2024
-
17
June 2024
Intervenant : | Anna Kazeykina |
Institution : | LMO (U. Paris-Saclay) |
Heure : | 14h00 - 14h50 |
Lieu : | 3L8 |
We consider a type of games regularized by the relative entropy, where the players' strategies are coupled through a random environment variable. We prove the existence and the uniqueness of equilibria of such games. We also demonstrate that the marginal laws of the corresponding mean-field Langevin systems can converge towards the games' equilibria in different settings. As an application, dynamic games can be treated as games on random environment when one treats the time horizon as the environment. In practice, our results can be applied to analysing the stochastic gradient descent algorithm for deep neural networks in the context of supervised learning as well as for the generative adversarial networks.