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

In recent years, advances in deep learning have resulted in a plethora ofsuccesses in the use of reinforcement learning (RL) to solve complex sequentialdecision tasks with high-dimensional inputs. However, existing systems lack thenecessary mechanisms to provide humans with a holistic view of theircompetence, presenting an impediment to their adoption, particularly incritical applications where the decisions an agent makes can have significantconsequences. Yet, existing RL-based systems are essentially competency-unawarein that they lack the necessary interpretation mechanisms to allow humanoperators to have an insightful, holistic view of their competency. Towardsmore explainable Deep RL (xDRL), we propose a new framework based on analysesof interestingness. Our tool provides various measures of RL agent competencestemming from interestingness analysis and is applicable to a wide range of RLalgorithms, natively supporting the popular RLLib toolkit. We showcase the useof our framework by applying the proposed pipeline in a set of scenarios ofvarying complexity. We empirically assess the capability of the approach inidentifying agent behavior patterns and competency-controlling conditions, andthe task elements mostly responsible for an agent's competence, based on globaland local analyses of interestingness. Overall, we show that our framework canprovide agent designers with insights about RL agent competence, both theircapabilities and limitations, enabling more informed decisions aboutinterventions, additional training, and other interactions in collaborativehuman-machine settings.

IxDRL: A Novel Explainable Deep Reinforcement Learning Toolkit based on Analyses of Interestingness