Can Model Compression Improve NLP Fairness

Xu, Guangxuan; Hu, Qingyuan

doi:10.48550/arxiv.2201.08542

Cited by 4 publications

(4 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Specifically regarding bias in the distillation or model compression setting, Xu and Hu (2022) report reduction in bias in contrast to our findings, although in a generation application. However, Gupta et al (2022) makes clear that biases from the training data can also be preserved or exacerbated in a similar distillation setting.…”

Section: Related Workcontrasting

confidence: 99%

Lost in Distillation: A Case Study in Toxicity Modeling

Chvasta¹,

Lees²,

Sorensen³

et al. 2022

Proceedings of the Sixth Workshop on Online Abuse and Harms (WOAH)

View full text Add to dashboard Cite

In an era of increasingly large pre-trained language models, knowledge distillation is a powerful tool for transferring information from a large model to a smaller one. In particular, distillation is of tremendous benefit when it comes to real-world constraints such as serving latency or serving at scale. However, a loss of robustness in language understanding may be hidden in the process and not immediately revealed when looking at high-level evaluation metrics. We investigate the hidden costs: what is "lost in distillation", especially in regards to identity-based model bias using the case study of toxicity modeling. With reproducible models using open source training sets, we investigate models distilled from a BERT teacher baseline. Using both open source and proprietary big data models, we investigate these hidden performance costs.

show abstract

Section: Related Workcontrasting

confidence: 99%

Lost in Distillation: A Case Study in Toxicity Modeling

Chvasta¹,

Lees²,

Sorensen³

et al. 2022

Proceedings of the Sixth Workshop on Online Abuse and Harms (WOAH)

View full text Add to dashboard Cite

show abstract

“…Additionally, [8] presented a regularisation procedure that aims at debiasing a language model by minimising the projection of encoder-trained embeddings onto a subspace that encodes gender. Similarly, [59] used model compression techniques, a type of regularisation techniques, to reduce toxicity and bias originally present in generative language models. The system proposed by [32] mitigates bias by employing counterfactual data augmentation, proving that modifying the training data works better than changing the actual geometry of the embeddings.…”

Section: Related Workmentioning

confidence: 99%

Improving Gender-Related Fairness in Sentence Encoders: A Semantics-Based Approach

2023

View full text Add to dashboard Cite

The ever-increasing number of systems based on semantic text analysis is making natural language understanding a fundamental task: embedding-based language models are used for a variety of applications, such as resume parsing or improving web search results. At the same time, despite their popularity and widespread use, concern is rapidly growing due to their display of social bias and lack of transparency. In particular, they exhibit a large amount of gender bias, favouring the consolidation of social stereotypes. Recently, sentence embeddings have been introduced as a novel and powerful technique to represent entire sentences as vectors. We propose a new metric to estimate gender bias in sentence embeddings, named bias score. Our solution leverages semantic importance of words and previous research on bias in word embeddings, and it is able to discern between neutral and biased gender information at sentence level. Experiments on a real-world dataset demonstrate that our novel metric can identify gender stereotyped sentences. Furthermore, we employ bias score to detect and then remove or compensate for the more stereotyped entries in text corpora used to train sentence encoders, improving their degree of fairness. Finally, we prove that models retrained on fairer corpora are less prone to make stereotypical associations compared to their original counterpart, while preserving accuracy in natural language understanding tasks. Additionally, we compare our experiments with traditional methods for reducing bias in embedding-based language models.

show abstract

“…Xu and Hu 37 showed that there is reduction of toxicity and bias in compressed GPT2 model compressed via Knowledge Distillation. In computer vision models, Quantization and Pruning methods have compromised fairness because performance of samples with under-represented features is sacrificed after compression ( 38 , 39 ).…”

Section: Bias In Compressed Modelsmentioning

confidence: 99%

Compressed Models for Co-reference Resolution: Enhancing Efficiency with Debiased Word Embeddings

Ioannides

Jadhav

Sharma

et al. 2023

Preprint

View full text Add to dashboard Cite

This work presents a comprehensive approach to reduce bias in word embedding vectors and evaluate the impact on various Natural Language Processing (NLP) tasks. Two GloVe variations (840B and 50) are debiased by identifying the gender direction in the word embedding space and then removing or reducing the gender component from the embeddings of target words, while preserving useful semantic information. Their gender bias is assessed through the Word Embedding Association Test. The performance of co-reference resolution and sentiment analysis models trained on both original and debiased embeddings is evaluated in terms of accuracy. A compressed co-reference resolution model is examined to gauge the effectiveness of debiasing techniques on resource-efficient models. To the best of the authors' knowledge, this is the first attempt to apply compression techniques to debiased models. By analyzing the context preservation of debiased embeddings using a Twitter misinformation dataset, this study contributes valuable insights into the practical implications of debiasing methods for real-world applications such as human profiling.

show abstract

Can Model Compression Improve NLP Fairness

Cited by 4 publications

References 14 publications

Lost in Distillation: A Case Study in Toxicity Modeling

Lost in Distillation: A Case Study in Toxicity Modeling

Improving Gender-Related Fairness in Sentence Encoders: A Semantics-Based Approach

Compressed Models for Co-reference Resolution: Enhancing Efficiency with Debiased Word Embeddings

Contact Info

Product

Resources

About