(45 minutes to learn)
The Dirichlet process is a stochastic process that defines a probability distribution over infinite-dimensional discrete distributions, meaning that a draw form a DP is itself a distribution (with a countably infinite number of parameters). Its name stems from the fact that the marginal of a DP for any finite partition is Dirichlet distributed. While the DP is often discussed alongside the Chinese Restaurant Process (CRP), the two are not the same entity. The DP is the de Finetti mixing measure for the CRP, meaning that sampling i.i.d. from a draw of a DP is equivalent to sequentially drawing samples from the CRP.
This concept has the prerequisites:
- Dirichlet distribution
- Chinese restaurant process (The Chinese restaurant process is one canonical interpretation of the Dirichlet process.)
Core resources (read/watch one of the following)
→ Graphical Models for Visual Object Recognition and Tracking (2006)
Erik Sudderth's Ph.D. thesis, which includes readable overviews of a variety of topics.
Location: section 2.5 Dirichlet Processes
Supplemental resources (the following are optional, but you may find them useful)
→ Dirichlet Process
- assumes some familiarity with measure theory
→ Bayesian Nonparametrics (2011)
Location: part 2, from 0:00 - 10:33
- DP is further discussed throughout the entire lecture
- The beta process is an analogue of the Dirichlet process which is useful for probabilistic models which represent binary attributes.
- Dirichlet diffusion trees are a hierarchical clustering model based on the same ideas as the Dirichlet process.
- create concept: shift + click on graph
- change concept title: shift + click on existing concept
- link together concepts: shift + click drag from one concept to another
- remove concept from graph: click on concept then press delete/backspace
- add associated content to concept: click the small circle that appears on the node when hovering over it
- other actions: use the icons in the upper right corner to optimize the graph placement, preview the graph, or download a json representation