Summary of Dcast: Diverse Class-aware Self-training Mitigates Selection Bias For Fairer Learning, by Yasin I. Tepeli et al.
DCAST: Diverse Class-Aware Self-Training Mitigates Selection Bias for Fairer Learning
by Yasin I. Tepeli, Joana P. Gonçalves
First submitted to arxiv on: 30 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computers and Society (cs.CY)
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 This paper tackles the issue of fairness in machine learning by addressing model bias towards individuals based on sensitive features like sex or age. The problem is often caused by uneven training data representation due to selection bias. The authors highlight that existing methods struggle to identify and mitigate this type of bias, which is prominent in complex high-dimensional data from fields like computer vision and molecular biomedicine. To address this challenge, the paper introduces two novel strategies: Diverse Class-Aware Self-Training (DCAST) and hierarchy bias. DCAST promotes sample diversity to counter confirmation bias while leveraging unlabeled samples for a more representative population. The authors demonstrate improved robustness of models learned with DCAST across 11 datasets, outperforming conventional self-training and domain adaptation techniques. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making sure machine learning models don’t treat people unfairly because of things like their age or gender. Right now, there’s a problem where training data isn’t representative of the whole population, which leads to biases in the model. The authors want to find ways to fix this and make the models more fair. They’re introducing two new methods: DCAST and hierarchy bias. DCAST helps by mixing up the training data to avoid confirmation bias and using extra information that isn’t labeled. This makes the models better at recognizing patterns and being fair. |
Keywords
» Artificial intelligence » Domain adaptation » Machine learning » Self training
