Multi-layer Relation Networks

TL;DR

This paper introduces a multi-layer relation network architecture that enhances relational reasoning capabilities, enabling the solution of complex tasks like all 20 bAbI QA dataset questions, surpassing previous state-of-the-art results.

Contribution

It proposes a novel multi-layer architecture for relation networks, allowing for deeper relational reasoning beyond pairwise interactions.

Findings

Successfully solves all 20 bAbI QA tasks with joint training.

Outperforms previous state-of-the-art results on the bAbI dataset.

Demonstrates the effectiveness of multi-layer relational reasoning.

Abstract

Relational Networks (RN) as introduced by Santoro et al. (2017) have demonstrated strong relational reasoning capabilities with a rather shallow architecture. Its single-layer design, however, only considers pairs of information objects, making it unsuitable for problems requiring reasoning across a higher number of facts. To overcome this limitation, we propose a multi-layer relation network architecture which enables successive refinements of relational information through multiple layers. We show that the increased depth allows for more complex relational reasoning by applying it to the bAbI 20 QA dataset, solving all 20 tasks with joint training and surpassing the state-of-the-art results.