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

Safe exploration presents a major challenge in reinforcement learning (RL):when active data collection requires deploying partially trained policies, wemust ensure that these policies avoid catastrophically unsafe regions, whilestill enabling trial and error learning. In this paper, we target the problemof safe exploration in RL by learning a conservative safety estimate ofenvironment states through a critic, and provably upper bound the likelihood ofcatastrophic failures at every training iteration. We theoreticallycharacterize the tradeoff between safety and policy improvement, show that thesafety constraints are likely to be satisfied with high probability duringtraining, derive provable convergence guarantees for our approach, which is noworse asymptotically than standard RL, and demonstrate the efficacy of theproposed approach on a suite of challenging navigation, manipulation, andlocomotion tasks. Empirically, we show that the proposed approach can achievecompetitive task performance while incurring significantly lower catastrophicfailure rates during training than prior methods. Videos are at this urlhttps://sites.google.com/view/conservative-safety-critics/home

Conservative Safety Critics for Exploration