2 papers
cs.LG2025
AMStraMGRAM: Adaptive Multi-cutoff Strategy Modification for ANaGRAM
Nilo Schwencke, Cyriaque Rousselot, Alena Shilova +1
Recent works have shown that natural gradient methods can significantly outperform standard optimizers when training physics-informed neural networks (PINNs). In this paper, we ana…
cs.LG2025
StaQ: a Finite Memory Approach to Discrete Action Policy Mirror Descent
Alena Shilova, Alex Davey, Brahim Driss +1
In Reinforcement Learning (RL), regularization with a Kullback-Leibler divergence that penalizes large deviations between successive policies has emerged as a popular tool both in…