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

This work explores learning agent-agnostic synthetic environments (SEs) forReinforcement Learning. SEs act as a proxy for target environments and allowagents to be trained more efficiently than when directly trained on the targetenvironment. We formulate this as a bi-level optimization problem and representan SE as a neural network. By using Natural Evolution Strategies and apopulation of SE parameter vectors, we train agents in the inner loop onevolving SEs while in the outer loop we use the performance on the target taskas a score for meta-updating the SE population. We show empirically that ourmethod is capable of learning SEs for two discrete-action-space tasks(CartPole-v0 and Acrobot-v1) that allow us to train agents more robustly andwith up to 60% fewer steps. Not only do we show in experiments with 4000evaluations that the SEs are robust against hyperparameter changes such as thelearning rate, batch sizes and network sizes, we also show that SEs trainedwith DDQN agents transfer in limited ways to a discrete-action-space version ofTD3 and very well to Dueling DDQN.

Learning Synthetic Environments for Reinforcement Learning with Evolution Strategies