Computing exact clustering posteriors with subset convolution

ABSTRACT An exponential-time exact algorithm is provided for the task of clustering n items of data into k clusters. Instead of seeking one partition, posterior probabilities are computed for summary statistics: the number of clusters and pairwise co-occurrence. The method is based on subset convolution and yields the posterior distribution for the number of clusters in O(n3n) operations or O(n32n) using fast subset convolution. Pairwise co-occurrence probabilities are then obtained in O(n32n) operations. This is considerably faster than exhaustive enumeration of all partitions.