Construction of Dependent Dirichlet Processes Based on Poisson Processes
Abstract
We present a novel method for constructing dependent Dirichlet processes. The approach exploits the intrinsic relationship between Dirichlet and Poisson pro- cesses in order to create a Markov chain of Dirichlet processes suitable for use as a prior over evolving mixture models. The method allows for the creation, re- moval, and location variation of component models over time while maintaining the property that the random measures are marginally DP distributed. Addition- ally, we derive a Gibbs sampling algorithm for model inference and test it on both synthetic and real data. Empirical results demonstrate that the approach is effec- tive in estimating dynamically varying mixture models.
Cite
Text
Lin et al. "Construction of Dependent Dirichlet Processes Based on Poisson Processes." Neural Information Processing Systems, 2010.Markdown
[Lin et al. "Construction of Dependent Dirichlet Processes Based on Poisson Processes." Neural Information Processing Systems, 2010.](https://mlanthology.org/neurips/2010/lin2010neurips-construction/)BibTeX
@inproceedings{lin2010neurips-construction,
title = {{Construction of Dependent Dirichlet Processes Based on Poisson Processes}},
author = {Lin, Dahua and Grimson, Eric and Fisher, John W.},
booktitle = {Neural Information Processing Systems},
year = {2010},
pages = {1396-1404},
url = {https://mlanthology.org/neurips/2010/lin2010neurips-construction/}
}