Gone With the Bits: Revealing Racial Bias in Low-Rate Neural Compression for Facial Images

Tian Qiu, Arjun Nichani, Rasta Tadayon, Haewon Jeong

Keywords: Fairness, Bias, Neural Compression, Phenotype Classification

TL;DR: We present a framework to evaluate racial bias in neural image compression models, showing that examining facial phenotype degradation reveals racial bias, and that a racially balanced training set reduces but doesn't fully eliminate this bias.

Abstract: Neural compression methods are gaining popularity due to their impressive rate-distortion performance and their ability to compress data to extremely small bitrates, below 0.1 bits per pixel (bpp). As deep learning architectures, these models are prone to bias during the training process, potentially leading to unfair outcomes for individuals in different groups. In this paper, we present a general, structured, scalable framework for evaluating bias in neural image compression models. Using this framework, we investigate racial bias in neural compression algorithms by analyzing 7 popular models and their variants. Through this investigation we first demonstrate that traditional distortion metrics are ineffective in capturing bias in neural compression models. Next, we highlight that racial bias is present in all neural compression models and can be captured by examining facial phenotype degradation in image reconstructions. Additionally, we reveal a task-dependent correlation between bias and model architecture. We then examine the relationship between bias and realism in the image reconstructions and demonstrate a trade-off across models. Finally, we show that utilizing a racially balanced training set can reduce bias but is not a sufficient bias mitigation strategy.

Primary Area: alignment, fairness, safety, privacy, and societal considerations

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 8943