Working Memory Networks: Augmenting Memory Networks with a Relational Reasoning Module

TL;DR

This paper introduces the Working Memory Network, a neural architecture that combines relational reasoning with efficient memory storage, achieving state-of-the-art results in question-answering tasks while reducing computational complexity.

Contribution

It presents a novel memory and reasoning module that retains relational reasoning capabilities of Relation Networks but with linear complexity, improving scalability.

Findings

Achieved less than 0.5% mean error on bAbI-10k dataset.

Solved all 20 tasks in the joint NLVR benchmark with an ensemble.

Reduced relational reasoning computational cost from quadratic to linear.

Abstract

During the last years, there has been a lot of interest in achieving some kind of complex reasoning using deep neural networks. To do that, models like Memory Networks (MemNNs) have combined external memory storages and attention mechanisms. These architectures, however, lack of more complex reasoning mechanisms that could allow, for instance, relational reasoning. Relation Networks (RNs), on the other hand, have shown outstanding results in relational reasoning tasks. Unfortunately, their computational cost grows quadratically with the number of memories, something prohibitive for larger problems. To solve these issues, we introduce the Working Memory Network, a MemNN architecture with a novel working memory storage and reasoning module. Our model retains the relational reasoning abilities of the RN while reducing its computational complexity from quadratic to linear. We tested our model on the text QA dataset bAbI and the visual QA dataset NLVR. In the jointly trained bAbI-10k, we set a new state-of-the-art, achieving a mean error of less than 0.5%. Moreover, a simple ensemble of two of our models solves all 20 tasks in the joint version of the benchmark.