Gradual Domain Adaptation via Manifold-Constrained Distributionally Robust Optimization

Neural Information Processing Systems 

The aim of this paper is to address the challenge of gradual domain adaptation within a class of manifold-constrained data distributions.