assignIdxPerClMBinEM

PjV1

PjV2

Assign final indices to data by maximum posterior value.

An implementation of Multi-View Clustering (Bickel and Scheffer, 2004). Documents are generated by drawing word values from a categorical distribution for each word, given the cluster. This means words are not counted (multinomial, as in the paper), but words take on different values from a finite set of values (categorical). Thus, it implements Mixture of Categoricals EM (as opposed to Mixture of Multinomials developed in the paper), and Spherical k-Means. The latter represents documents as vectors in the categorical space.

assignIdxPerClMBinEM: Assign final indices to data by maximum posterior value.

Description

Usage

Arguments

Value