CTL++: Evaluating Generalization on Never-Seen Compositional Patterns of Known Functions, and Compatibility of Neural Representations

Róbert Csordás, Kazuki Irie, Juergen Schmidhuber


Abstract
Well-designed diagnostic tasks have played a key role in studying the failure of neural nets (NNs) to generalize systematically. Famous examples include SCAN and Compositional Table Lookup (CTL). Here we introduce CTL++, a new diagnostic dataset based on compositions of unary symbolic functions. While the original CTL is used to test length generalization or productivity, CTL++ is designed to test systematicity of NNs, that is, their capability to generalize to unseen compositions of known functions. CTL++ splits functions into groups and tests performance on group elements composed in a way not seen during training. We show that recent CTL-solving Transformer variants fail on CTL++. The simplicity of the task design allows for fine-grained control of task difficulty, as well as many insightful analyses. For example, we measure how much overlap between groups is needed by tested NNs for learning to compose. We also visualize how learned symbol representations in outputs of functions from different groups are compatible in case of success but not in case of failure. These results provide insights into failure cases reported on more complex compositions in the natural language domain. Our code is public.
Anthology ID:
2022.emnlp-main.662
Volume:
Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
9758–9767
Language:
URL:
https://aclanthology.org/2022.emnlp-main.662
DOI:
10.18653/v1/2022.emnlp-main.662
Bibkey:
Cite (ACL):
Róbert Csordás, Kazuki Irie, and Juergen Schmidhuber. 2022. CTL++: Evaluating Generalization on Never-Seen Compositional Patterns of Known Functions, and Compatibility of Neural Representations. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 9758–9767, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
CTL++: Evaluating Generalization on Never-Seen Compositional Patterns of Known Functions, and Compatibility of Neural Representations (Csordás et al., EMNLP 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.emnlp-main.662.pdf