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

Offline reinforcement learning aims to train a policy on a pre-recorded andfixed dataset without any additional environment interactions. There are twomajor challenges in this setting: (1) extrapolation error caused byapproximating the value of state-action pairs not well-covered by the trainingdata and (2) distributional shift between behavior and inference policies. Oneway to tackle these problems is to induce conservatism - i.e., keeping thelearned policies closer to the behavioral ones. To achieve this, we build uponrecent works on learning policies in latent action spaces and use a specialform of Normalizing Flows for constructing a generative model, which we use asa conservative action encoder. This Normalizing Flows action encoder ispre-trained in a supervised manner on the offline dataset, and then anadditional policy model - controller in the latent space - is trained viareinforcement learning. This approach avoids querying actions outside of thetraining dataset and therefore does not require additional regularization forout-of-dataset actions. We evaluate our method on various locomotion andnavigation tasks, demonstrating that our approach outperforms recently proposedalgorithms with generative action models on a large portion of datasets.

Let Offline RL Flow: Training Conservative Agents in the Latent Space of Normalizing Flows