Software

balancing   Code and experiments for the paper “The Benefits of Balance: From Information Projections to Variance Reduction” (NeurIPS ‘24).

drago   Code and experiments for the paper “Drago: Primal-Dual Coupled Variance Reduction for Faster Distributionally Robust Optimization” (NeurIPS ‘24).

deshift   A package for stochastic distributionally robust learning in PyTorch with support for data distributed workflows.

prospect   Code and experiments for the paper “Distributionally Robust Optimization with Bias and Variance Reduction” (ICLR ‘24 Spotlight).

lerm   Code and experiments for the paper “Stochastic Algorithms for Spectral Risk Measures” (AISTATS ‘23).

statcluster-quickstart   A quickstart guide to Python work on the University of Washington Statistics Department computing cluster (or any CPU cluster, really).