Understanding Early Word Learning in Situated Artificial Agents

TL;DR

This paper investigates how neural network-based agents learn to understand and interpret single-word instructions in a simulated 3D environment, inspired by developmental psychology, and introduces a new visualization method for semantic representations.

Contribution

It provides insights into the mechanisms of grounded language learning in neural agents and proposes a novel visualization technique for semantic representations.

Findings

Models can overcome learning obstacles without prior knowledge

Human-like biases and effects emerge in the agent under certain conditions

A new visualization method reveals semantic structure in neural representations

Abstract

Neural network-based systems can now learn to locate the referents of words and phrases in images, answer questions about visual scenes, and execute symbolic instructions as first-person actors in partially-observable worlds. To achieve this so-called grounded language learning, models must overcome challenges that infants face when learning their first words. While it is notable that models with no meaningful prior knowledge overcome these obstacles, researchers currently lack a clear understanding of how they do so, a problem that we attempt to address in this paper. For maximum control and generality, we focus on a simple neural network-based language learning agent, trained via policy-gradient methods, which can interpret single-word instructions in a simulated 3D world. Whilst the goal is not to explicitly model infant word learning, we take inspiration from experimental paradigms in developmental psychology and apply some of these to the artificial agent, exploring the conditions under which established human biases and learning effects emerge. We further propose a novel method for visualising semantic representations in the agent.