Dynamic Clustering via Asymptotics of the Dependent Dirichlet Process\n Mixture

2013/05/28 by Trevor Campbell, Campbell, Trevor, Miao Liu +7 · 1 citation
Computer Science · Mathematics · #Bayesian Methods and Mixture Models #Advanced Clustering Algorithms Research #Statistical Methods and Bayesian Inference

paper · pdf · doi:10.48550/arxiv.1305.6659

Abstract

This paper presents a novel algorithm, based upon the dependent Dirichlet\nprocess mixture model (DDPMM), for clustering batch-sequential data containing\nan unknown number of evolving clusters. The algorithm is derived via a\nlow-variance asymptotic analysis of the Gibbs sampling algorithm for the DDPMM,\nand provides a hard clustering with convergence guarantees similar to those of\nthe k-means algorithm. Empirical results from a synthetic test with moving\nGaussian clusters and a test with real ADS-B aircraft trajectory data\ndemonstrate that the algorithm requires orders of magnitude less computational\ntime than contemporary probabilistic and hard clustering algorithms, while\nproviding higher accuracy on the examined datasets.\n

Citations

• Least squares quantization in PCM

Cited by

• Inference for multiple object tracking: A Bayesian nonparametric approach

Related

• Revisiting k-means: New Algorithms via Bayesian Nonparametrics
• Sampling in Dirichlet Process Mixture Models for Clustering Streaming Data
• Dynamic Clustering Algorithms via Small-Variance Analysis of Markov\n Chain Mixture Models
• Dirichlet Process Mixtures of Generalized Mallows Models
• Beta-Product Poisson-Dirichlet Processes