Summary of Himoe: Heterogeneity-informed Mixture-of-experts For Fair Spatial-temporal Forecasting, by Shaohan Yu et al.
HiMoE: Heterogeneity-Informed Mixture-of-Experts for Fair Spatial-Temporal Forecasting
by Shaohan Yu, Pan Deng, Yu Zhao, Junting Liu, Zi’ang Wang
First submitted to arxiv on: 30 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
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 novel approach to achieve fair prediction performance across nodes in spatial-temporal forecasting tasks. The existing models focus on improving overall accuracy but neglect uniformity in predictions, which is crucial for validity and reliability of forecasting outcomes. To address this issue, the authors introduce Heterogeneity-informed Mixture-of-Experts (HiMoE) that combines HiGCN and NMoE to model spatial dependencies and allocate prediction tasks to suitable experts. The paper also proposes fairness-aware loss and evaluation functions to optimize the model for both fairness and accuracy. Experimental results on four datasets demonstrate the state-of-the-art performance of HiMoE, outperforming the best baseline with at least 9.22% in all metrics. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about making predictions that are fair and accurate across different places and times. Right now, most models try to make overall predictions better, but they don’t worry much about how well they do at each specific place or time. This can be a problem because we need to know that our predictions are reliable and good for everyone. To solve this issue, the authors created a new way to predict using something called Heterogeneity-informed Mixture-of-Experts (HiMoE). HiMoE is made up of two parts: one that looks at how places are connected and another that decides which part should do the predicting. The paper also came up with ways to make sure our predictions are fair and good, not just accurate. They tested it on four different real-world datasets and found that their new method works much better than previous ones. |
Keywords
» Artificial intelligence » Mixture of experts
