Zero-Shot Compositional Policy Learning via Language Grounding
This addresses the generalization challenge in AI agents for domain adaptation, though it is incremental as it builds on existing language-guided techniques.
The paper tackles the problem of reinforcement and imitation learning agents failing to generalize beyond training environments by proposing a zero-shot compositional policy learning task, and it finds that a multi-modal fusion method with attention for language grounding improves generalization across varied dynamics, with evidence from experiments on the new BabyAI++ platform.
Despite recent breakthroughs in reinforcement learning (RL) and imitation learning (IL), existing algorithms fail to generalize beyond the training environments. In reality, humans can adapt to new tasks quickly by leveraging prior knowledge about the world such as language descriptions. To facilitate the research on language-guided agents with domain adaption, we propose a novel zero-shot compositional policy learning task, where the environments are characterized as a composition of different attributes. Since there are no public environments supporting this study, we introduce a new research platform BabyAI++ in which the dynamics of environments are disentangled from visual appearance. At each episode, BabyAI++ provides varied vision-dynamics combinations along with corresponding descriptive texts. To evaluate the adaption capability of learned agents, a set of vision-dynamics pairings are held-out for testing on BabyAI++. Unsurprisingly, we find that current language-guided RL/IL techniques overfit to the training environments and suffer from a huge performance drop when facing unseen combinations. In response, we propose a multi-modal fusion method with an attention mechanism to perform visual language-grounding. Extensive experiments show strong evidence that language grounding is able to improve the generalization of agents across environments with varied dynamics.