Summary of Fair-tat: Improving Model Fairness Using Targeted Adversarial Training, by Tejaswini Medi et al.
FAIR-TAT: Improving Model Fairness Using Targeted Adversarial Training
by Tejaswini Medi, Steffen Jung, Margret Keuper
First submitted to arxiv on: 30 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
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 paper presents a novel approach called Fair Targeted Adversarial Training (FAIR-TAT) that addresses the issues of robustness and fairness in deep neural networks against adversarial attacks and common corruptions. The authors highlight the limitations of existing approaches, such as Adversarial Training (AT), which often sacrifice model fairness for increased robustness. They demonstrate that using targeted adversarial attacks during training can lead to more favorable trade-offs between robustness and fairness. The paper shows that FAIR-TAT outperforms state-of-the-art models in terms of both robustness and fairness, making it a promising solution for enhancing the resilience of deep learning models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about making sure artificial intelligence models are not too good at fooling themselves with fake pictures or words. Right now, some ways to make AI more resistant to these tricks actually make it less fair in its predictions. This paper introduces a new approach called Fair Targeted Adversarial Training that can help AI be both strong and fair. It uses special types of fake pictures to train the AI, which helps it make better decisions. |
Keywords
* Artificial intelligence * Deep learning
