English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Action-constrained reinforcement learning (RL) is a widely-used approach invarious real-world applications, such as scheduling in networked systems withresource constraints and control of a robot with kinematic constraints. Whilethe existing projection-based approaches ensure zero constraint violation, theycould suffer from the zero-gradient problem due to the tight coupling of thepolicy gradient and the projection, which results in sample-inefficienttraining and slow convergence. To tackle this issue, we propose a learningalgorithm that decouples the action constraints from the policy parameterupdate by leveraging state-wise Frank-Wolfe and a regression-based policyupdate scheme. Moreover, we show that the proposed algorithm enjoys convergenceand policy improvement properties in the tabular case as well as generalizesthe popular DDPG algorithm for action-constrained RL in the general case.Through experiments, we demonstrate that the proposed algorithm significantlyoutperforms the benchmark methods on a variety of control tasks.

Escaping from Zero Gradient: Revisiting Action-Constrained Reinforcement Learning via Frank-Wolfe Policy Optimization