Recovery of Sparse Probability Measures via Convex Programming

Neural Information Processing Systems 

We consider the problem of cardinality penalized optimization of a convex function over the probability simplex with additional convex constraints.