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

Reinforcement learning (RL) involves performing exploratory actions in anunknown system. This can place a learning agent in dangerous and potentiallycatastrophic system states. Current approaches for tackling safe learning in RLsimultaneously trade-off safe exploration and task fulfillment. In this paper,we introduce a new generation of RL solvers that learn to minimise safetyviolations while maximising the task reward to the extent that can be toleratedby the safe policy. Our approach introduces a novel two-player framework forsafe RL called Distributive Exploration Safety Training Algorithm (DESTA). Thecore of DESTA is a game between two adaptive agents: Safety Agent that isdelegated the task of minimising safety violations and Task Agent whose goal isto maximise the environment reward. Specifically, Safety Agent can selectivelytake control of the system at any given point to prevent safety violationswhile Task Agent is free to execute its policy at any other states. Thisframework enables Safety Agent to learn to take actions at certain states thatminimise future safety violations, both during training and testing time, whileTask Agent performs actions that maximise the task performance everywhere else.Theoretically, we prove that DESTA converges to stable points enabling safetyviolations of pretrained policies to be minimised. Empirically, we show DESTA'sability to augment the safety of existing policies and secondly, construct safeRL policies when the Task Agent and Safety Agent are trained concurrently. Wedemonstrate DESTA's superior performance against leading RL methods in LunarLander and Frozen Lake from OpenAI gym.

DESTA: A Framework for Safe Reinforcement Learning with Markov Games of Intervention