MilaNLP at SemEval-2023 Task 10: Ensembling Domain-Adapted and Regularized Pretrained Language Models for Robust Sexism Detection

Amanda Cercas Curry, Giuseppe Attanasio, Debora Nozza, Dirk Hovy


Abstract
We present the system proposed by the MilaNLP team for the Explainable Detection of Online Sexism (EDOS) shared task. We propose an ensemble modeling approach to combine different classifiers trained with domain adaptation objectives and standard fine-tuning. Our results show that the ensemble is more robust than individual models and that regularized models generate more “conservative” predictions, mitigating the effects of lexical overfitting.However, our error analysis also finds that many of the misclassified instances are debatable, raising questions about the objective annotatability of hate speech data.
Anthology ID:
2023.semeval-1.285
Volume:
Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023)
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Atul Kr. Ojha, A. Seza Doğruöz, Giovanni Da San Martino, Harish Tayyar Madabushi, Ritesh Kumar, Elisa Sartori
Venue:
SemEval
SIG:
SIGLEX
Publisher:
Association for Computational Linguistics
Note:
Pages:
2067–2074
Language:
URL:
https://aclanthology.org/2023.semeval-1.285
DOI:
10.18653/v1/2023.semeval-1.285
Bibkey:
Cite (ACL):
Amanda Cercas Curry, Giuseppe Attanasio, Debora Nozza, and Dirk Hovy. 2023. MilaNLP at SemEval-2023 Task 10: Ensembling Domain-Adapted and Regularized Pretrained Language Models for Robust Sexism Detection. In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 2067–2074, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
MilaNLP at SemEval-2023 Task 10: Ensembling Domain-Adapted and Regularized Pretrained Language Models for Robust Sexism Detection (Cercas Curry et al., SemEval 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.semeval-1.285.pdf
Video:
 https://aclanthology.org/2023.semeval-1.285.mp4