Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
102 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Robust Evaluation Measures for Evaluating Social Biases in Masked Language Models (2401.11601v1)

Published 21 Jan 2024 in cs.CL

Abstract: Many evaluation measures are used to evaluate social biases in masked LLMs (MLMs). However, we find that these previously proposed evaluation measures are lacking robustness in scenarios with limited datasets. This is because these measures are obtained by comparing the pseudo-log-likelihood (PLL) scores of the stereotypical and anti-stereotypical samples using an indicator function. The disadvantage is the limited mining of the PLL score sets without capturing its distributional information. In this paper, we represent a PLL score set as a Gaussian distribution and use Kullback Leibler (KL) divergence and Jensen Shannon (JS) divergence to construct evaluation measures for the distributions of stereotypical and anti-stereotypical PLL scores. Experimental results on the publicly available datasets StereoSet (SS) and CrowS-Pairs (CP) show that our proposed measures are significantly more robust and interpretable than those proposed previously.

View on arXiv
Definition Search Book Streamline Icon: https://streamlinehq.com
References (33)
  1. Stereotyping Norwegian salmon: An inventory of pitfalls in fairness benchmark datasets. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), 1004–1015.
  2. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. Advances in neural information processing systems, 29.
  3. Language models are few-shot learners. Advances in neural information processing systems, 33: 1877–1901.
  4. Semantics derived automatically from language corpora contain human-like biases. Science, 356(6334): 183–186.
  5. Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Setting. In Proceedings of the Conference on Fairness, Accountability, and Transparency, FAT* ’19, 120–128. New York, NY, USA: Association for Computing Machinery. ISBN 9781450361255.
  6. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), 4171–4186. Minneapolis, Minnesota: Association for Computational Linguistics.
  7. A new metric for probability distributions. IEEE Transactions on Information theory, 49(7): 1858–1860.
  8. Jensen-Shannon divergence and Hilbert space embedding. In International Symposium onInformation Theory, 2004. ISIT 2004. Proceedings., 31. IEEE.
  9. Universal Language Model Fine-tuning for Text Classification. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 328–339. Melbourne, Australia: Association for Computational Linguistics.
  10. Unmasking the mask–evaluating social biases in masked language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, 11954–11962.
  11. Sample estimate of the entropy of a random vector. Problemy Peredachi Informatsii, 23(2): 9–16.
  12. Estimating mutual information. Physical review E, 69(6): 066138.
  13. On information and sufficiency. The annals of mathematical statistics, 22(1): 79–86.
  14. ALBERT: A Lite BERT for Self-supervised Learning of Language Representations. In International Conference on Learning Representations.
  15. Mining Effective Features Using Quantum Entropy for Humor Recognition. In Findings of the Association for Computational Linguistics: EACL 2023, 2048–2053. Dubrovnik, Croatia: Association for Computational Linguistics.
  16. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
  17. On Measuring Social Biases in Sentence Encoders. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), 622–628.
  18. Efficient inference through cascades of weighted tree transducers. In Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, 1058–1066.
  19. StereoSet: Measuring stereotypical bias in pretrained language models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), 5356–5371.
  20. CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 1953–1967.
  21. Parzen, E. 1962. On estimation of a probability density function and mode. The annals of mathematical statistics, 33(3): 1065–1076.
  22. Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), 1532–1543.
  23. Deep Contextualized Word Representations. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), 2227–2237. New Orleans, Louisiana: Association for Computational Linguistics.
  24. Language models are unsupervised multitask learners. OpenAI blog, 1(8): 9.
  25. Rosenblatt, M. 1956. Remarks on some nonparametric estimates of a density function. The annals of mathematical statistics, 832–837.
  26. Ross, B. C. 2014. Mutual information between discrete and continuous data sets. PloS one, 9(2): e87357.
  27. Rozado, D. 2020. Wide range screening of algorithmic bias in word embedding models using large sentiment lexicons reveals underreported bias types. PloS one, 15(4): e0231189.
  28. Masked Language Model Scoring. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 2699–2712. Online: Association for Computational Linguistics.
  29. Transfer learning strategies for solar power forecasting under data scarcity. Scientific Reports, 12(1): 14643.
  30. An analysis of variance test for normality (complete samples)†. Biometrika, 52(3-4): 591–611.
  31. BERT has a Mouth, and It Must Speak: BERT as a Markov Random Field Language Model. In Proceedings of the Workshop on Methods for Optimizing and Evaluating Neural Language Generation, 30–36. Minneapolis, Minnesota: Association for Computational Linguistics.
  32. Measuring and reducing gendered correlations in pre-trained models. arXiv preprint arXiv:2010.06032.
  33. Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers), 15–20.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (1)
  1. Yang Liu (2253 papers)
Citations (2)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now