BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning
This addresses the problem of poor data efficiency in training AI agents to understand language instructions, which is incremental as it builds on existing methods with a new benchmark.
The paper introduces BabyAI, a platform with 19 difficulty levels to study grounded language learning, and provides evidence that current deep learning methods lack sample efficiency for compositional language acquisition, estimating high human involvement for training.
Allowing humans to interactively train artificial agents to understand language instructions is desirable for both practical and scientific reasons, but given the poor data efficiency of the current learning methods, this goal may require substantial research efforts. Here, we introduce the BabyAI research platform to support investigations towards including humans in the loop for grounded language learning. The BabyAI platform comprises an extensible suite of 19 levels of increasing difficulty. The levels gradually lead the agent towards acquiring a combinatorially rich synthetic language which is a proper subset of English. The platform also provides a heuristic expert agent for the purpose of simulating a human teacher. We report baseline results and estimate the amount of human involvement that would be required to train a neural network-based agent on some of the BabyAI levels. We put forward strong evidence that current deep learning methods are not yet sufficiently sample efficient when it comes to learning a language with compositional properties.