Equivariant Transduction through Invariant Alignment

TL;DR

This paper introduces a novel group-equivariant neural network with an invariant alignment mechanism that enhances compositional generalization in NLP tasks, outperforming previous models on the SCAN benchmark.

Contribution

The paper proposes a new group-equivariant architecture with an invariant alignment mechanism, improving equivariance and performance on the SCAN task compared to prior models.

Findings

The new architecture develops stronger equivariance properties.

It outperforms previous group-equivariant networks on SCAN.

The approach highlights the importance of theoretical analysis of equivariance in neural models.

Abstract

The ability to generalize compositionally is key to understanding the potentially infinite number of sentences that can be constructed in a human language from only a finite number of words. Investigating whether NLP models possess this ability has been a topic of interest: SCAN (Lake and Baroni, 2018) is one task specifically proposed to test for this property. Previous work has achieved impressive empirical results using a group-equivariant neural network that naturally encodes a useful inductive bias for SCAN (Gordon et al., 2020). Inspired by this, we introduce a novel group-equivariant architecture that incorporates a group-invariant hard alignment mechanism. We find that our network's structure allows it to develop stronger equivariance properties than existing group-equivariant approaches. We additionally find that it outperforms previous group-equivariant networks empirically on the SCAN task. Our results suggest that integrating group-equivariance into a variety of neural architectures is a potentially fruitful avenue of research, and demonstrate the value of careful analysis of the theoretical properties of such architectures.