Bayesian Agglomerative Clustering with Coalescents

Abstract

We introduce a new Bayesian model for hierarchical clustering based on a prior over trees called Kingman's coalescent. We develop novel greedy and sequential Monte Carlo inferences which operate in a bottom-up agglomerative fashion. We show experimentally the superiority of our algorithms over others, and demonstrate our approach in document clustering and phylolinguistics.

NIPS 2008

References in corpus (1)

• A Bayesian Model for Discovering Typological Implications

Cited by in corpus (17)

• Bayesian Multitask Learning with Latent Hierarchies
• Bayesian Rose Trees
• The Infinite Hierarchical Factor Regression Model
• Bayesian Multitask Learning with Latent Hierarchies
• Divide-and-Conquer with Sequential Monte Carlo
• Multiresolution Gaussian Processes
• Ancestor Sampling for Particle Gibbs
• Efficient Continuous-Time Markov Chain Estimation
• Pitman-Yor Diffusion Trees
• Top-down particle filtering for Bayesian decision trees
• Tree-Structured Stick Breaking Processes for Hierarchical Data
• Unfolding Latent Tree Structures using 4th Order Tensors
• Interactive Bayesian Hierarchical Clustering
• Efficient hierarchical clustering for continuous data
• Unsupervised Embedding of Hierarchical Structure in Euclidean Space
• Inferring Hierarchical Mixture Structures: A Bayesian Nonparametric Approach
• Constructing a Family Tree of Ten Indo-European Languages with Delexicalized Cross-linguistic Transfer Patterns