3D-Learning: Diffusion-Augmented Distributionally Robust Decision-Focused Learning
Jiaqi Wen, Lei Fan, Jianyi Yang* · 2026
Read 3D-Learning: Diffusion-Augmented Distributionally Robust Decision-Focused Learning by Jiaqi Wen, Lei Fan, Jianyi Yang* (2026) 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-2602-02943.txt.
Description
Machines that predict first and optimize later can fail badly on unusual inputs, and Jiaqi Wen and 2 co-authors address that risk in a 2026 paper. Such pipelines forecast demand or context, then feed the forecast into a decision. Out-of-distribution test cases can cause large prediction errors and poor decisions.
Their framework, called 3D-Learning, trains predictors for worst-case decision quality. A diffusion model searches for the worst plausible data distribution, staying consistent with real data. The method balances average and worst-case performance. Tests use a language-model resource provisioning task, where computing must be allocated ahead of demand. The approach beat robust optimization and data augmentation baselines on out-of-distribution generalization.
Catalog record
| CREATOR | Jiaqi Wen, Lei Fan, Jianyi Yang* |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-02943