Abstract by Spencer Newcomb
An Improved Merge-Split Sampler for Dirichlet Process Mixture Models
The Dirichlet process mixture (DPM) model is a popular Bayesian nonparametric model. Markov chain Monte Carlo (MCMC) algorithms are often used to sample from the posterior distribution. Several MCMC algorithms exist, all of which theoretically sample from the posterior distribution in the limit. However, in finite sampling, some algorithms do not adequately explore valleys of low probability density, getting “stuck” in local modes – a problem that is accentuated in higher dimensions. In this presentation, we provide a simple overview of the DPM model and review commonly-used MCMC approaches. We propose a more efficient sampling algorithm based on the splitting and merging of components. Borrowing ideas from sequential importance sampling (SIS), we expect that our algorithm will have better computational efficiency in terms of distributional convergence and scaling than other available split-merge algorithms.