Recently, Awasthi et al. introduced an SDP relaxation of the -means problem in . In this work, we consider a random model for the data points in which balls of unit radius are deterministically distributed throughout , and then in each ball, points are drawn according to a common rotationally invariant probability distribution. For any fixed ball configuration and probability distribution, we prove that the SDP relaxation of the -means problem exactly recovers these planted clusters with probability provided the distance between any two of the ball centers is , where is an explicit function of the configuration of the ball centers, and can be arbitrarily small when is large.
View on arXiv