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

Policy optimization methods have shown great promise in solving complexreinforcement and imitation learning tasks. While model-free methods arebroadly applicable, they often require many samples to optimize complexpolicies. Model-based methods greatly improve sample-efficiency but at the costof poor generalization, requiring a carefully handcrafted model of the systemdynamics for each task. Recently, hybrid methods have been successful intrading off applicability for improved sample-complexity. However, these havebeen limited to continuous action spaces. In this work, we present a new hybridmethod based on an approximation of the dynamics as an expectation over thenext state under the current policy. This relaxation allows us to derive anovel hybrid policy gradient estimator, combining score function and pathwisederivative estimators, that is applicable to discrete action spaces. We showsignificant gains in sample complexity, ranging between $1.7$ and $25\times$,when learning parameterized policies on Cart Pole, Acrobot, Mountain Car andHand Mass. Our method is applicable to both discrete and continuous actionspaces, when competing pathwise methods are limited to the latter.

Deterministic Policy Optimization by Combining Pathwise and Score Function Estimators for Discrete Action Spaces