1: \begin{abstract}

2: Normalized random measures (NRMs) provide a broad class of discrete random measures that are 

3: often used as priors for Bayesian nonparametric models. Dirichlet process is a well-known example of NRMs.

4: Most of posterior inference methods for NRM mixture models rely on MCMC methods since they

5: are easy to implement and their convergence is well studied.

6: However, MCMC often suffers from slow convergence when the acceptance rate is low.

7: Tree-based inference is an alternative deterministic posterior inference method, where

8: Bayesian hierarchical clustering (BHC) or incremental Bayesian hierarchical clustering (IBHC) have been

9: developed for DP or NRM mixture (NRMM) models, respectively.

10: Although IBHC is a promising method for posterior inference for NRMM models due to

11: its efficiency and applicability to online inference, its convergence is not guaranteed

12: since it uses heuristics that simply selects the best solution after multiple trials are made.

13: In this paper, we present a hybrid inference algorithm for NRMM models, which combines

14: the merits of both MCMC and IBHC. Trees built by IBHC outlines partitions of data, which guides

15: Metropolis-Hastings procedure to employ appropriate proposals. Inheriting the nature of MCMC,

16: our tree-guided MCMC (tgMCMC) is guaranteed to converge, and enjoys the fast convergence 

17: thanks to the effective proposals guided by trees. Experiments on both synthetic

18: and real-world datasets demonstrate the benefit of our method. 

19: \end{abstract}

20: