Reviews: Learning Nonsymmetric Determinantal Point Processes

This paper studies determinantal point processes (DPP) with non-symmetric kernels. Most of the machine learning literature on DPP assumes symmetric kernels, and the prior work that studies non-symmetric kernels have assumed a quite restricted class of non-symmetric kernels. The novelty of this paper is in proposing the learning algorithm for a fairly general class of non-symmetric kernels. The proposed approach assumes a particular representation of non-symmetric kernels. This representation follows from two known results in a rather straightforward manner, as I also summarize in "1.