Work

Efficient Wasserstein Natural Gradients for Reinforcement Learning

Year: 2020

Type: preprint

Abstract: A novel optimization approach is proposed for application to policy gradient methods and evolution strategies for reinforcement learning (RL). The procedure uses a computationally efficient Wasserstein natural gradient (WNG) descent that takes advantage of the geometry induced by a Wasserstein penalty to speed optimization. This method follows the recent theme in RL of including a divergence penalty in the objective to establish a trust region. Experiments on challenging tasks demonstrate improvements in both computational cost and performance over advanced baselines.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.2010.05380

Authors: Moskovitz, Ted, Arbel, Michael, Huszar, Ferenc, Gretton, Arthur

Institutions: Gatsby Computational Neuroscience Unit, University College London, University of Cambridge

Language: English

Topic: Reinforcement Learning in Robotics

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W3092698374

Cites: 25

Cited by: 1

Related to: 20