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

While the difficulty of reinforcement learning problems is typically relatedto the complexity of their state spaces, Abstraction proposes that solutionsoften lie in simpler underlying latent spaces. Prior works have focused onlearning either a continuous or dense abstraction, or require a human toprovide one. Information-dense representations capture features irrelevant forsolving tasks, and continuous spaces can struggle to represent discreteobjects. In this work we automatically learn a sparse discrete abstraction ofthe underlying environment. We do so using a simple end-to-end trainable modelbased on the successor representation and max-entropy regularization. Wedescribe an algorithm to apply our model, named Discrete State-ActionAbstraction (DSAA), which computes an action abstraction in the form oftemporally extended actions, i.e., Options, to transition between discreteabstract states. Empirically, we demonstrate the effects of differentexploration schemes on our resulting abstraction, and show that it is efficientfor solving downstream tasks.

Discrete State-Action Abstraction via the Successor Representation