Summary of Overcoming Representation Bias in Fairness-aware Data Repair Using Optimal Transport, by Abigail Langbridge and Anthony Quinn and Robert Shorten
Overcoming Representation Bias in Fairness-Aware data Repair using Optimal Transport
by Abigail Langbridge, Anthony Quinn, Robert Shorten
First submitted to arxiv on: 3 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computers and Society (cs.CY); Statistics Theory (math.ST)
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 proposes a new approach to optimal transport (OT) for transforming data distributions while ensuring fairness, addressing two significant limitations of previous methods. The authors develop a Bayesian nonparametric stopping rule to learn attribute-labelled components of the data distribution, which enables OT-optimal quantization operators to repair archival data. A novel definition of fair distributional targets is also introduced, along with quantifiers that allow for trading off fairness against damage in the transformed data. The proposed scheme demonstrates excellent performance in simulated and benchmark datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper uses a special kind of math called optimal transport to make sure data is fair. Right now, this process doesn’t work well when there’s not enough information about certain groups or when working with old data that we’ve never seen before. The authors create a new way to fix these problems by using something called Bayesian nonparametric stopping rules. This helps them create special operators that can repair old data and make sure it’s fair. They also come up with a new way to define what “fair” means in this context, which is important because sometimes we might have to sacrifice some fairness to get better results. |
Keywords
* Artificial intelligence * Quantization
