Grounding Language to Entities and Dynamics for Generalization in Reinforcement Learning
arXiv:2101.07393
Abstract
We investigate the use of natural language to drive the generalization of control policies and introduce the new multi-task environment Messenger with free-form text manuals describing the environment dynamics. Unlike previous work, Messenger does not assume prior knowledge connecting text and state observations the control policy must simultaneously ground the game manual to entity symbols and dynamics in the environment. We develop a new model, EMMA (Entity Mapper with Multi-modal Attention) which uses an entity-conditioned attention module that allows for selective focus over relevant descriptions in the manual for each entity in the environment. EMMA is end-to-end differentiable and learns a latent grounding of entities and dynamics from text to observations using only environment rewards. EMMA achieves successful zero-shot generalization to unseen games with new dynamics, obtaining a 40% higher win rate compared to multiple baselines. However, win rate on the hardest stage of Messenger remains low (10%), demonstrating the need for additional work in this direction.
arXiv commentAccepted to ICML 2021. Note author list and name changes from previous version
References in corpus (10)
- Modular Multitask Reinforcement Learning with Policy Sketches
- Grounded Language Learning in a Simulated 3D World
- The NetHack Learning Environment
- Grounded Language Learning Fast and Slow
- Human Instruction-Following with Deep Reinforcement Learning via Transfer-Learning from Text
- Learning to Map Natural Language Instructions to Physical Quadcopter Control using Simulated Flight
- Grounding Language to Entities and Dynamics for Generalization in Reinforcement Learning
- Alignment-based compositional semantics for instruction following
- Are You Looking? Grounding to Multiple Modalities in Vision-and-Language Navigation
- Guiding Policies with Language via Meta-Learning