Summary of Fast & Fair: Efficient Second-order Robust Optimization For Fairness in Machine Learning, by Allen Minch et al.
Fast & Fair: Efficient Second-Order Robust Optimization for Fairness in Machine Learning
by Allen Minch, Hung Anh Vu, Anne Marie Warren
First submitted to arxiv on: 4 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computers and Society (cs.CY); Numerical Analysis (math.NA)
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 As machine learning educators strive to create fairer Deep Neural Networks (DNNs), researchers have turned to adversarial training techniques to mitigate the inherent bias in these networks. DNNs can perpetuate biases rooted in sensitive attributes like race and gender, leading to life-altering consequences, as seen in facial recognition software used for suspect arrests. The proposed robust optimization problem has been demonstrated to improve fairness in various datasets, both synthetic and real-world, utilizing an affine linear model. By leveraging second-order information, this approach efficiently finds a solution, outperforming purely first-order methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Researchers are working on making Deep Neural Networks (DNNs) fairer by using special training techniques called adversarial training. Right now, DNNs can have biases that affect important decisions. For example, facial recognition software used to catch suspects might be biased against certain groups of people. This project shows how a new way of solving a math problem can make DNNs fairer in different types of data. |
Keywords
* Artificial intelligence * Machine learning * Optimization
