Disentangling and Integrating Relational and Sensory Information in Transformer Architectures

TL;DR

This paper introduces the Dual Attention Transformer (DAT), an extension of the Transformer architecture that explicitly separates and processes sensory and relational information, improving performance on relational reasoning tasks.

Contribution

The paper proposes a novel architectural extension, DAT, with separate attention mechanisms for sensory and relational information, addressing a key limitation of standard Transformers.

Findings

DAT outperforms standard Transformers on relational benchmarks

Explicit relational attention improves data and parameter efficiency

The approach benefits language and visual processing tasks

Abstract

Relational reasoning is a central component of generally intelligent systems, enabling robust and data-efficient inductive generalization. Recent empirical evidence shows that many existing neural architectures, including Transformers, struggle with tasks requiring relational reasoning. In this work, we distinguish between two types of information: sensory information about the properties of individual objects, and relational information about the relationships between objects. While neural attention provides a powerful mechanism for controlling the flow of sensory information between objects, the Transformer lacks an explicit computational mechanism for routing and processing relational information. To address this limitation, we propose an architectural extension of the Transformer framework that we call the Dual Attention Transformer (DAT), featuring two distinct attention mechanisms: sensory attention for directing the flow of sensory information, and a novel relational attention mechanism for directing the flow of relational information. We empirically evaluate DAT on a diverse set of tasks ranging from synthetic relational benchmarks to complex real-world tasks such as language modeling and visual processing. Our results demonstrate that integrating explicit relational computational mechanisms into the Transformer architecture leads to significant performance gains in terms of data efficiency and parameter efficiency.