Summary of Fairness in Survival Analysis with Distributionally Robust Optimization, by Shu Hu et al.
Fairness in Survival Analysis with Distributionally Robust Optimization
by Shu Hu, George H. Chen
First submitted to arxiv on: 31 Aug 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: 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 A novel approach for ensuring fairness in survival analysis models is proposed, utilizing distributionally robust optimization (DRO) to minimize worst-case errors across all subpopulations with a user-specified probability. This method can be applied to many existing survival analysis models, converting them into fair and accurate variants without requiring sensitive attribute specification. By applying recent DRO developments to survival analysis, the approach addresses the technical hurdle of decomposing loss functions that commonly involve ranking or similarity score calculations. Sample splitting strategies are employed to overcome this challenge, demonstrating the conversion of various existing models, including Cox, DeepSurv, DeepHit, and SODEN, into fair variants. Theoretical guarantees are established for finite-sample convergence, showing improved fairness metrics without significant accuracy drops compared to existing regularization techniques. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Fairness in survival analysis is important, but it can be tricky to achieve. Researchers have come up with a new way to make sure survival models are fair and accurate. They use something called distributionally robust optimization (DRO) to minimize worst-case errors across different groups of people. This approach doesn’t require specifying which attributes or features should be treated as sensitive, making it easier to use. The researchers tested this method on several existing models and showed that it can improve fairness without sacrificing accuracy. |
Keywords
» Artificial intelligence » Optimization » Probability » Regularization
