Non-Stationary Learning of Neural Networks with Automatic Soft Parameter Reset
Abstract
Neural networks are traditionally trained under the assumption that data come from a stationary distribution. However, settings which violate this assumption are becoming more popular; examples include supervised learning under distributional shifts, reinforcement learning, continual learning and non-stationary contextual bandits. In this work we introduce a novel learning approach that automatically models and adapts to non-stationarity, via an Ornstein-Uhlenbeck process with an adaptive drift parameter. The adaptive drift tends to draw the parameters towards the initialisation distribution, so the approach can be understood as a form of soft parameter reset. We show empirically that our approach performs well in non-stationary supervised and off-policy reinforcement learning settings.
Cite
@article{arxiv.2411.04034,
title = {Non-Stationary Learning of Neural Networks with Automatic Soft Parameter Reset},
author = {Alexandre Galashov and Michalis K. Titsias and András György and Clare Lyle and Razvan Pascanu and Yee Whye Teh and Maneesh Sahani},
journal= {arXiv preprint arXiv:2411.04034},
year = {2024}
}