Make Every Example Count: On the Stability and Utility of Self-Influence for Learning from Noisy NLP Datasets

Irina Bejan, Artem Sokolov, Katja Filippova


Abstract
Increasingly larger datasets have become a standard ingredient to advancing the state-of-the-art in NLP. However, data quality might have already become the bottleneck to unlock further gains. Given the diversity and the sizes of modern datasets, standard data filtering is not straight-forward to apply, because of the multifacetedness of the harmful data and elusiveness of filtering rules that would generalize across multiple tasks. We study the fitness of task-agnostic self-influence scores of training examples for data cleaning, analyze their efficacy in capturing naturally occurring outliers, and investigate to what extent self-influence based data cleaning can improve downstream performance in machine translation, question answering and text classification, building up on recent approaches to self-influence calculation and automated curriculum learning.
Anthology ID:
2023.emnlp-main.625
Volume:
Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
Month:
December
Year:
2023
Address:
Singapore
Editors:
Houda Bouamor, Juan Pino, Kalika Bali
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
10107–10121
Language:
URL:
https://aclanthology.org/2023.emnlp-main.625
DOI:
10.18653/v1/2023.emnlp-main.625
Bibkey:
Cite (ACL):
Irina Bejan, Artem Sokolov, and Katja Filippova. 2023. Make Every Example Count: On the Stability and Utility of Self-Influence for Learning from Noisy NLP Datasets. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 10107–10121, Singapore. Association for Computational Linguistics.
Cite (Informal):
Make Every Example Count: On the Stability and Utility of Self-Influence for Learning from Noisy NLP Datasets (Bejan et al., EMNLP 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/ingest-2024-clasp/2023.emnlp-main.625.pdf
Video:
 https://preview.aclanthology.org/ingest-2024-clasp/2023.emnlp-main.625.mp4