Conformal-DRO paper builds ambiguity sets from nested conformal regions for robust decisions
chaumian · x · 2026-09-13
The arXiv paper "Conformal-DRO: Distributionally Robust Optimization with Conformalized Ambiguity Set" (Luhao Zhang, Shixiang Zhu) tackles latent distributional heterogeneity in data-driven DRO:
- Setting: each instance has an unobserved law but contributes one observation, so uncertainty persists even when the mixture law is known.
- Method: Conformal-DRO uses nested conformal regions to build an ambiguity set for the future latent law; under exchangeability it covers the law with probability ≥ 1−α in finite samples, without estimating the latent laws.
- Solving: the conformal path induces a data-driven transport geometry; the worst-case problem reduces to a finite linear program with sparse adversarial solutions, yielding a finite-sample certificate on expected cost.
More from Research
- FAccT Researcher: Safety Evals Ignore Rich Measurement Work Beyond Benchmarks — evijit · 2026-09-13
- Brynjolfsson Pushes GDP-B: Digital Goods Create Value Yet GDP May Fall — erikbryn · 2026-09-13
- Lab insider explains the internal-external gap on AI progress and Astra scare — MajmudarAdam · 2026-09-13
- NiiVue wrapper ecosystem brings interactive neuroimaging visualization to VS Code, Jupyter, R and the web — pshrink · 2026-09-13
- SlopCodeBench draws praise as a new benchmark for multi-turn code degradation — cedric_chee · 2026-09-13
- 33-author paper maps the road to recursive self-improvement, reviewing 72 AI teams — huybery · 2026-09-13