Summary of Fairstg: Countering Performance Heterogeneity Via Collaborative Sample-level Optimization, by Gengyu Lin et al.
FairSTG: Countering performance heterogeneity via collaborative sample-level optimization
by Gengyu Lin, Zhengyang Zhou, Qihe Huang, Kuo Yang, Shifen Cheng, Yang Wang
First submitted to arxiv on: 19 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 addresses the issue of unfairness in mobile computing techniques used to empower smart cities. Existing research has focused on achieving accurate predictions on overall datasets but neglected significant performance heterogeneity across samples, leading to degraded model practicality and real-world risks. The proposed Fairness-aware framework for SpatioTemporal Graph learning (FairSTG) aims to fix this gap by exploiting advantages of well-learned samples to challenging ones through collaborative mix-up. FairSTG consists of a spatiotemporal feature extractor, representation enhancement, and fairness objectives to suppress sample-level performance heterogeneity. The framework is evaluated on four spatiotemporal datasets, demonstrating significant fairness improvements while maintaining comparable forecasting accuracy. Case studies show that FairSTG can counter both spatial and temporal performance heterogeneity by sample-level retrieval and compensation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research paper tries to make smart cities’ systems more fair. Right now, these systems often work better for some areas than others, which is a problem. The researchers propose a new way to train these systems so that they are more fair. They call it FairSTG. This new method helps by sharing information between different parts of the city and making sure all areas get treated equally. The researchers tested this method on four real-world datasets and found that it makes the systems more fair without sacrificing their accuracy. |
Keywords
* Artificial intelligence * Spatiotemporal
