Inductive Biases for Zero-shot Systematic Generalization in Language-informed Reinforcement Learning

TL;DR

This paper introduces architecture-level inductive biases in language-informed reinforcement learning, leveraging neural production systems and memory to enhance systematic generalization and sample efficiency in the BabyAI environment.

Contribution

It proposes a novel architecture with inductive biases, neural production systems, and memory mechanisms to improve generalization and efficiency in language-informed RL.

Findings

Significant improvement in systematic generalization in BabyAI

Enhanced sample efficiency over previous models

Ablation study clarifies the impact of each technique

Abstract

Sample efficiency and systematic generalization are two long-standing challenges in reinforcement learning. Previous studies have shown that involving natural language along with other observation modalities can improve generalization and sample efficiency due to its compositional and open-ended nature. However, to transfer these properties of language to the decision-making process, it is necessary to establish a proper language grounding mechanism. One approach to this problem is applying inductive biases to extract fine-grained and informative representations from the observations, which makes them more connectable to the language units. We provide architecture-level inductive biases for modularity and sparsity mainly based on Neural Production Systems (NPS). Alongside NPS, we assign a central role to memory in our architecture. It can be seen as a high-level information aggregator which feeds policy/value heads with comprehensive information and simultaneously guides selective attention in NPS through attentional feedback. Our results in the BabyAI environment suggest that the proposed model's systematic generalization and sample efficiency are improved significantly compared to previous models. An extensive ablation study on variants of the proposed method is conducted, and the effectiveness of each employed technique on generalization, sample efficiency, and training stability is specified.