University of Cambridge > > Machine Learning @ CUED > The Block Diagonal Infinite Hidden Markov Model

The Block Diagonal Infinite Hidden Markov Model

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Zoubin Ghahramani.

The Infinite Hidden Markov Model (IHMM) is a nonparametric Bayesian discrete time series model. By maintaining a posterior distribution over transition and emission models for a countably infinite number of hidden states, the IHMM flexibly accommodates data for which the “right” number of such states is not known.

The Block Diagonal Infinite Hidden Markov Model (BD-IHMM) extends the IHMM to target circumstances where the data alternate over time among a collection of distinct behavioral regimes, or “sub-behaviors”. In ordinary HMMs, alternation like this arises from transition matrices with nearly block-diagonal structure, and the BD-IHMM’s prior induces similar forms in inferred dynamics, here too with flexibility in the number of blocks. By inferring a hidden state sequence for the data, the model also associates all parts of the data with a sub-behavior, a sequence-clustering capability that may also have useful analogs in settings besides time series.

This talk will present the BD-IHMM, describe inference techniques for the model, show recent results (these involve the Nintendo Wii and ‘90s USA alt-rock sensation Collective Soul), and discuss extensions and new directions under investigation now.

This talk is part of the Machine Learning @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity