Summary of A Self-supervised Learning Pipeline For Demographically Fair Facial Attribute Classification, by Sreeraj Ramachandran and Ajita Rattani
A Self-Supervised Learning Pipeline for Demographically Fair Facial Attribute Classification
by Sreeraj Ramachandran, Ajita Rattani
First submitted to arxiv on: 14 Jul 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The abstract discusses demographic bias in automated facial attribute classification. The proposed mitigation techniques rely on supervised learning, which requires large amounts of labeled data. However, labeled data is limited, labor-intensive to annotate, and can perpetuate human bias. Self-supervised learning (SSL) leverages freely available unlabeled data for scalability and generalizability, but may introduce biases and performance degradation. This paper proposes a fully self-supervised pipeline for demographically fair facial attribute classifiers using pseudolabeling, diverse data curation, and meta-learning-based weighted contrastive learning. The method outperforms existing SSL approaches on the FairFace and CelebA datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research shows that computers can unfairly judge people’s faces. To fix this problem, the team proposes a new way to train computer models using only unseen data, without needing labeled pictures of different faces. This approach is fairer than previous methods because it doesn’t rely on human bias. The new method does better than old ones in tests with large datasets. |
Keywords
» Artificial intelligence » Classification » Meta learning » Self supervised » Supervised
