{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/exact-slice-sampler-for-hierarchical","title":"Exact slice sampler for Hierarchical Dirichlet Processes","arxiv_id":"1903.08829","date":"2019-03-21","proceeding":null,"authors":["Arash A. Amini","Marina Paez","Lizhen Lin","Zahra S. Razaee"],"abstract":"We propose an exact slice sampler for Hierarchical Dirichlet process (HDP)\nand its associated mixture models (Teh et al., 2006). Although there are\nexisting MCMC algorithms for sampling from the HDP, a slice sampler has been\nmissing from the literature. Slice sampling is well-known for its desirable\nproperties including its fast mixing and its natural potential for\nparallelization. On the other hand, the hierarchical nature of HDPs poses\nchallenges to adopting a full-fledged slice sampler that automatically\ntruncates all the infinite measures involved without ad-hoc modifications. In\nthis work, we adopt the powerful idea of Bayesian variable augmentation to\naddress this challenge. By introducing new latent variables, we obtain a full\nfactorization of the joint distribution that is suitable for slice sampling.\nOur algorithm has several appealing features such as (1) fast mixing; (2)\nremaining exact while allowing natural truncation of the underlying\ninfinite-dimensional measures, as in (Kalli et al., 2011), resulting in updates\nof only a finite number of necessary atoms and weights in each iteration; and\n(3) being naturally suited to parallel implementations. The underlying\nprinciple for joint factorization of the full likelihood is simple and can be\napplied to many other settings, such as designing sampling algorithms for\ngeneral dependent Dirichlet process (DDP) models.","url_abs":"http://arxiv.org/abs/1903.08829v1","url_pdf":"http://arxiv.org/pdf/1903.08829v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"exact-slice-sampler-for-hierarchical","repo_url":"https://github.com/aaamini/hdpslicer","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"exact-slice-sampler-for-hierarchical","repo_url":"https://github.com/das-snigdha/blockedhdp","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1903.08829","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}