Material Detail

Time Dependent Stick Breaking Processes

Time Dependent Stick Breaking Processes

This video was recorded at 25th International Conference on Machine Learning (ICML), Helsinki 2008. The stick breaking construction of the Dirichlet process has been a popular starting point for many dependent nonparametric processes. This talk considers a temporal version of the Order-Based Dependent Dirichlet Process, which can be extended to more general stick-breaking marginal processes. Interestingly, the simplest constructions lead to marginal Dirichlet and Poisson-Dirichlet processes. Usefully, the first process also has a "Chinese restaurant"-type representation which will be described. Applications to time-dependent mixture modelling will be presented.

Quality

  • User Rating
  • Comments
  • Learning Exercises
  • Bookmark Collections
  • Course ePortfolios
  • Accessibility Info

More about this material

Comments

Log in to participate in the discussions or sign up if you are not already a MERLOT member.