Compositional Generalisation with Structured Reordering and Fertility Layers

Matthias Lindemann, Alexander Koller, Ivan Titov


Abstract
Seq2seq models have been shown to struggle with compositional generalisation, i.e. generalising to new and potentially more complex structures than seen during training. Taking inspiration from grammar-based models that excel at compositional generalisation, we present a flexible end-to-end differentiable neural model that composes two structural operations: a fertility step, which we introduce in this work, and a reordering step based on previous work (Wang et al., 2021). To ensure differentiability, we use the expected value of each step, which we compute using dynamic programming. Our model outperforms seq2seq models by a wide margin on challenging compositional splits of realistic semantic parsing tasks that require generalisation to longer examples. It also compares favourably to other models targeting compositional generalisation.
Anthology ID:
2023.eacl-main.159
Volume:
Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics
Month:
May
Year:
2023
Address:
Dubrovnik, Croatia
Editors:
Andreas Vlachos, Isabelle Augenstein
Venue:
EACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2172–2186
Language:
URL:
https://aclanthology.org/2023.eacl-main.159
DOI:
10.18653/v1/2023.eacl-main.159
Bibkey:
Cite (ACL):
Matthias Lindemann, Alexander Koller, and Ivan Titov. 2023. Compositional Generalisation with Structured Reordering and Fertility Layers. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2172–2186, Dubrovnik, Croatia. Association for Computational Linguistics.
Cite (Informal):
Compositional Generalisation with Structured Reordering and Fertility Layers (Lindemann et al., EACL 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.eacl-main.159.pdf
Video:
 https://aclanthology.org/2023.eacl-main.159.mp4