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 Reinforcement Learning (RL) approaches lead to significant achievementsin a variety of areas in recent history, natural language tasks remained mostlyunaffected, due to the compositional and combinatorial nature that makes themnotoriously hard to optimize. With the emerging field of Text-Based Games(TBGs), researchers try to bridge this gap. Inspired by the success of RLalgorithms on Atari games, the idea is to develop new methods in a restrictedgame world and then gradually move to more complex environments. Previous workin the area of TBGs has mainly focused on solving individual games. We,however, consider the task of designing an agent that not just succeeds in asingle game, but performs well across a whole family of games, sharing the sametheme. In this work, we present our deep RL agent--LeDeepChef--that showsgeneralization capabilities to never-before-seen games of the same family withdifferent environments and task descriptions. The agent participated inMicrosoft Research's "First TextWorld Problems: A Language and ReinforcementLearning Challenge" and outperformed all but one competitor on the final testset. The games from the challenge all share the same theme, namely cooking in amodern house environment, but differ significantly in the arrangement of therooms, the presented objects, and the specific goal (recipe to cook). To buildan agent that achieves high scores across a whole family of games, we use anactor-critic framework and prune the action-space by using ideas fromhierarchical reinforcement learning and a specialized module trained on arecipe database.

LeDeepChef: Deep Reinforcement Learning Agent for Families of Text-Based Games