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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

A popular framework for enforcing safe actions in Reinforcement Learning (RL)is Constrained RL, where trajectory based constraints on expected cost (orother cost measures) are employed to enforce safety and more importantly theseconstraints are enforced while maximizing expected reward. Most recentapproaches for solving Constrained RL convert the trajectory based costconstraint into a surrogate problem that can be solved using minormodifications to RL methods. A key drawback with such approaches is an over orunderestimation of the cost constraint at each state. Therefore, we provide anapproach that does not modify the trajectory based cost constraint and insteadimitates ``good'' trajectories and avoids ``bad'' trajectories generated fromincrementally improving policies. We employ an oracle that utilizes a rewardthreshold (which is varied with learning) and the overall cost constraint tolabel trajectories as ``good'' or ``bad''. A key advantage of our approach isthat we are able to work from any starting policy or set of trajectories andimprove on it. In an exhaustive set of experiments, we demonstrate that ourapproach is able to outperform top benchmark approaches for solving ConstrainedRL problems, with respect to expected cost, CVaR cost, or even unknown costconstraints.

Imitate the Good and Avoid the Bad: An Incremental Approach to Safe Reinforcement Learning