Distributionally robust linear regression through the lens of adversarial training

Name: Distributionally robust linear regression through the lens of adversarial training
Uploaded: 2026-10-04T21:19:58.017263+00:00
Duration: 270 s
Description: This work introduces Wasserstein Distributionally Robust Optimization (DRO) for linear regression, unifying results from Lasso and adversarial training. The method provides robust risk bounds with fast error rates of O(n⁻¹) under certain conditions, proving its pivotal property holds across various ambiguity sets.

This work introduces Wasserstein Distributionally Robust Optimization (DRO) for linear regression, unifying results from Lasso and adversarial training. The method provides robust risk bounds with fast error rates of O(n⁻¹) under certain conditions, proving its pivotal property holds across various ambiguity sets.

Open the full episode