Autoclass — A Bayesian Approach to Classification

摘要: We describe a Bayesian approach to the unsupervised discovery of classes in set cases, sometimes called finite mixture separation or clustering. The main difference between clustering and our is that we search for “best” class descriptions rather than grouping cases themselves. terms probability distribution density functions, locally maximal posterior parameters. rate classifications with an approximate function w.r.t. data, obtained by marginalizing over all Approximation necessitated computational complexity joint probability, marginalization local maxima parameter space. This rating allows direct comparison alternate functions differ number and/or individual functions.