ELLA: Exploration through Learned Language Abstraction
This addresses the challenge of sample inefficiency in language-based reinforcement learning for human-AI collaboration, but it is incremental as it builds on prior reward shaping and abstraction methods.
The paper tackles the problem of training agents to understand language instructions in sparse-reward environments by introducing ELLA, a reward shaping approach that boosts sample efficiency by correlating high-level instructions with low-level constituents, showing gains in sample efficiency on BabyAI environments.
Building agents capable of understanding language instructions is critical to effective and robust human-AI collaboration. Recent work focuses on training these agents via reinforcement learning in environments with synthetic language; however, instructions often define long-horizon, sparse-reward tasks, and learning policies requires many episodes of experience. We introduce ELLA: Exploration through Learned Language Abstraction, a reward shaping approach geared towards boosting sample efficiency in sparse reward environments by correlating high-level instructions with simpler low-level constituents. ELLA has two key elements: 1) A termination classifier that identifies when agents complete low-level instructions, and 2) A relevance classifier that correlates low-level instructions with success on high-level tasks. We learn the termination classifier offline from pairs of instructions and terminal states. Notably, in departure from prior work in language and abstraction, we learn the relevance classifier online, without relying on an explicit decomposition of high-level instructions to low-level instructions. On a suite of complex BabyAI environments with varying instruction complexities and reward sparsity, ELLA shows gains in sample efficiency relative to language-based shaping and traditional RL methods.