English

Equivariant Transduction through Invariant Alignment

Computation and Language 2022-09-23 v1 Machine Learning

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.

Keywords

Cite

@article{arxiv.2209.10926,
  title  = {Equivariant Transduction through Invariant Alignment},
  author = {Jennifer C. White and Ryan Cotterell},
  journal= {arXiv preprint arXiv:2209.10926},
  year   = {2022}
}

Comments

Accepted at COLING 2022

R2 v1 2026-06-28T01:53:22.077Z