Wasserstein-regularized conformal prediction under general distribution shift
Rui Xu, Sihong Xie, Chao Chen, Yue Sun, Parvathinathan Venkitasubramaniam · 2025
Read Wasserstein-regularized conformal prediction under general distribution shift by Rui Xu, Sihong Xie, Chao Chen, Yue Sun, Parvathinathan Venkitasubramaniam (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-13430.txt.
Description
Rui Xu and co-authors present a 2025 arXiv paper on keeping prediction guarantees honest when the world shifts. Conformal prediction promises that its answer sets cover the truth at a set rate, but only when test data match training data. The authors bound the coverage shortfall with Wasserstein distance, a measure sensitive to how far the data moved at a given confidence level.
Their analysis separates the damage done by shifting inputs from shifting input-answer links. A repair algorithm then reweights samples and regularizes the learned representation to shrink that bound. On six datasets it cut coverage gaps to about three percent while shrinking prediction sets by over a third.
Catalog record
| CREATOR | Rui Xu, Sihong Xie, Chao Chen, Yue Sun, Parvathinathan Venkitasubramaniam |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-13430