Clustering with Bregman divergences

摘要: Given a Bregman divergence dφ and a random variable X that takes values in a finite set X={xi}⊂ Ω⊂ Rd, where Ω is convex. Let ν denote the probability measure X induces. We pick a simple codebook C={s} and the Bregman information Iφ (X) is defined as the optimal distortion-rate function