Bayesian nonparametrics
Online resources:
- Collected resources by Peter Orbanz
- Video lectures by Tamara Broderick and Michael Jordan
- Lecture I: nice intuitions and buildup for Beta distribution, Dirichlet distribution (low and high dimensional), Dirichlet process stick-breaking and mixture model
Methods
Mixture models
- Dirichlet process
- The most basic nonparametric mixture model
- Aka: Chinese restaurant process, Beta stick-breaking process, GEM model (Griffiths, Engen and McCloskey)
- Neal, 2000: Markov chain sampling methods for Dirichlet process mixture models (doi, pdf, tech report )
- Pitman-Yor process, aka two-parameter Poisson-Dirichlet process
- Generalizes Dirichlet process to allow cluster sizes with power law distribution
- Teh, 2006: A hierarchical Bayesian language model based on Pitman-Yor processes (doi, tech report )
- Buntine & Hutter, 2010: A Bayesian view of the Poisson-Dirichlet process (arxiv)
- Hierarchical Dirichlet process
Latent variable models
- Indian buffet process
- Griffiths & Ghahramani, 2011: The Indian buffer process: An introduction and review (pdf)
Regression
Literature
Notes and surveys
- Orbanz, 2013: Lecture notes on Bayesian nonparametrics (pdf)
- Gershman and Blei, 2012: A tutorial on Bayesian nonparametric methods (doi,
arxiv, pdf)
- Friendly introduction to mixture models (CRP) and latent variable models (IBP)
- Ghahramani, 2012: Bayesian nonparametrics and the probabilistic approach to modeling (doi, pdf)
- Sudderth, 2006, PhD thesis: Graphical models for visual object recognition and tracking (uri , pdf), Sec 2.5: Dirichlet processes