f-Divergence Distributionally Robust Optimization

This-divergence chapter is devoted to f-divergence distributionally robust optimization, where the ambiguity set consists of the probability distributions whose f-divergence to the empirical probability distribution is within a given threshold. An overview of f-divergence is presented first to give a sense of what the f-divergence ambiguity set is, including the definitions, important properties, and the relationships between different types of f-divergences. Then the f-divergence ambiguity set is formally defined, followed by the reformulation methods of the corresponding worst-case expectations and robust chance constraints.