Summary of Robust Conformal Prediction Under Distribution Shift Via Physics-informed Structural Causal Model, by Rui Xu et al.
Robust Conformal Prediction under Distribution Shift via Physics-Informed Structural Causal Model
by Rui Xu, Yue Sun, Chao Chen, Parv Venkitasubramaniam, Sihong Xie
First submitted to arxiv on: 22 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
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 In this research paper, the authors focus on improving the reliability of machine learning models by addressing uncertainty. Specifically, they explore conformal prediction (CP), which involves predicting a set of possible outputs that covers the true label with at least 1-alpha confidence. The authors show that CP can be used to guarantee coverage on test data even when the input distributions differ between calibration and test datasets. However, they also highlight the importance of measuring and minimizing the coverage loss under distributional shift. To address this issue, they propose a physics-informed structural causal model (PI-SCM) that upper bounds the coverage difference at all levels using cumulative density functions and Wasserstein distance. The authors validate their approach on traffic speed prediction and epidemic spread tasks using multiple real-world datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Machine learning models are used to make predictions, but sometimes these predictions can be wrong. To fix this problem, researchers developed a method called conformal prediction (CP). CP is like a box that tries to cover the correct answer with a certain level of confidence. The authors of this paper want to improve CP by making it more reliable and less likely to make mistakes. They do this by using information from physics to help CP work better, especially when the data being used to make predictions is different from the data used to train the model. This can be important because in real life, data often changes and we need our models to adapt. The authors tested their new method on two tasks: predicting traffic speed and tracking an epidemic. They used real-world data for these tasks. |
Keywords
» Artificial intelligence » Machine learning » Tracking
