Abstract
We introduce our probabilistic robustness rewarded data optimization (PRoDO) approach as a framework to enhance the model’s generalization power by selecting training data that optimizes our probabilistic robustness metrics. We use proximal policy optimization (PPO) reinforcement learning to approximately solve the computationally intractable training subset selection problem. The PPO’s reward is defined as our (𝛼,𝜖, 𝛾)-Robustness that measures performance consistency over multiple domains by simulating unknown test sets in real-world scenarios using a leaving-one-out strategy. We demonstrate that our PRoDO effectively filters data that lead to significantly higher prediction accuracy and robustness on unknown-domain test sets. Our experiments achieve up to +17.2% increase of accuracy (+25.5% relatively) in sentiment analysis, and -28.05 decrease of perplexity (-32.1% relatively) in language modeling.In addition, our probabilistic (𝛼,𝜖, 𝛾)-Robustness definition serves as an evaluation metric with higher levels of agreement with human annotations than typical performance-based metrics.- Anthology ID:
- 2023.eacl-main.214
- 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:
- 2950–2959
- Language:
- URL:
- https://aclanthology.org/2023.eacl-main.214
- DOI:
- 10.18653/v1/2023.eacl-main.214
- Cite (ACL):
- Yu Yu, Abdul Rafae Khan, Shahram Khadivi, and Jia Xu. 2023. Probabilistic Robustness for Data Filtering. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2950–2959, Dubrovnik, Croatia. Association for Computational Linguistics.
- Cite (Informal):
- Probabilistic Robustness for Data Filtering (Yu et al., EACL 2023)
- PDF:
- https://preview.aclanthology.org/landing_page/2023.eacl-main.214.pdf