A common goal in statistics and machine learning is to learn models that can perform well against distributional shifts, such as latent heterogeneous subpopulations, unknown covariate shifts, or unmodeled temporal effects. We develop and analyze a distributionally robust stochastic optimization (DRO) framework that learns a model that provides good performance against perturbations to the data-generating distribution. We give a convex optimization formulation for the problem, providing several convergence guarantees. We prove finite-sample minimax upper and lower bounds, showing that distributinoal robustness sometimes comes at a cost in convergence rates. We give limit theorems for the learned parameters, where we fully specify the limiting distribution so that confidence intervals can be computed. On real tasks including generalizing to unknown subpopulations, fine-grained recognition, and providing good tail performance, the distributionally robust approach often exhibits improved performance.

本研究提出了一种分布鲁棒的随机优化框架，利用凸形式化来解决学习模型受到数据生成分布扰动的问题，并通过多项收敛性保准来证明模型的可靠性，同时也得出了极限定理及有关泛化到未知人群、精细化认知等真实任务的证据。

通过分布式稳健优化实现统一性能学习模型