 |
Center for Nonlinear Studies |
Clustering is a common tool in various sciences to analyze data in order to find similarities and discrepancies within data. Given numerical data, the goal of clustering is to identify regions, called clusters, of the data points such that points within the same cluster are closer to each other than to those within different clusters. The clustering of the data points by their proximity corresponds to similarity of the data represented. We present a clustering algorithm, derived using ideas from Bayesian inference and statistical physics, to identify such regions. We compare our algorithm to other common clustering algorithms.
Host: Matt Hastings