University of Cambridge > Talks.cam > Microsoft Research Cambridge, public talks > Probabilistic models for time-series with different underlying dynamics regimes with application to robot imitation learning

Probabilistic models for time-series with different underlying dynamics regimes with application to robot imitation learning

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

If you have a question about this talk, please contact Microsoft Research Cambridge Talks Admins.

In this talk I will describe several probabilistic models for analysing time-series containing different underlying dynamics regimes, and discuss approximation schemes for dealing with intractable inference. I will first present a Bayesian approach to switching and mixtures of linear dynamical systems (LDS) for incorporating prior domain knowledge and enforcing a sparse parametrisation. I will then introduce an extension of the switching LDS with regime-duration-distribution and time-warping modeling for extracting repeated occurrences of basic shapes. I will show how these models can be applied to robot imitation learning.

This talk is part of the Microsoft Research Cambridge, public talks series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2019 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity