Discovering the Compositional Structure of Vector Representations with Role Learning Networks

TL;DR

This paper introduces ROLE, a novel analysis technique that reveals how recurrent neural networks implicitly learn symbolic, compositional structures in vector representations, explaining their success on compositional tasks.

Contribution

The paper presents ROLE, a new method to uncover symbolic structures in neural network embeddings, demonstrating how RNNs implicitly learn compositional representations.

Findings

RNNs converge to solutions with implicit symbolic structure

Manipulating embeddings based on this structure alters outputs as predicted

The discovered structure closely matches the encodings of trained seq2seq models

Abstract

How can neural networks perform so well on compositional tasks even though they lack explicit compositional representations? We use a novel analysis technique called ROLE to show that recurrent neural networks perform well on such tasks by converging to solutions which implicitly represent symbolic structure. This method uncovers a symbolic structure which, when properly embedded in vector space, closely approximates the encodings of a standard seq2seq network trained to perform the compositional SCAN task. We verify the causal importance of the discovered symbolic structure by showing that, when we systematically manipulate hidden embeddings based on this symbolic structure, the model's output is changed in the way predicted by our analysis.