Reducing over-clustering via the powered Chinese restaurant process
This addresses the issue of over-clustering in mixture models for researchers and practitioners dealing with large datasets, though it appears incremental as it modifies an existing process.
The paper tackled the problem of Dirichlet process mixture models producing many small clusters, which hampers interpretability and efficiency, by proposing a powered Chinese restaurant process to penalize over-clustering, as demonstrated through simulations and datasets like MNIST and Old Faithful Geyser.
Dirichlet process mixture (DPM) models tend to produce many small clusters regardless of whether they are needed to accurately characterize the data - this is particularly true for large data sets. However, interpretability, parsimony, data storage and communication costs all are hampered by having overly many clusters. We propose a powered Chinese restaurant process to limit this kind of problem and penalize over clustering. The method is illustrated using some simulation examples and data with large and small sample size including MNIST and the Old Faithful Geyser data.