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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent work has described neural-network-based agents that are trained withreinforcement learning (RL) to execute language-like commands in simulatedworlds, as a step towards an intelligent agent or robot that can be instructedby human users. However, the optimisation of multi-goal motor policies via deepRL from scratch requires many episodes of experience. Consequently,instruction-following with deep RL typically involves language generated fromtemplates (by an environment simulator), which does not reflect the varied orambiguous expressions of real users. Here, we propose a conceptually simplemethod for training instruction-following agents with deep RL that are robustto natural human instructions. By applying our method with a state-of-the-artpre-trained text-based language model (BERT), on tasks requiring agents toidentify and position everyday objects relative to other objects in anaturalistic 3D simulated room, we demonstrate substantially-above-chancezero-shot transfer from synthetic template commands to natural instructionsgiven by humans. Our approach is a general recipe for training any deepRL-based system to interface with human users, and bridges the gap between tworesearch directions of notable recent success: agent-centric motor behavior andtext-based representation learning.

Human Instruction-Following with Deep Reinforcement Learning via Transfer-Learning from Text