Some Practical Reflections on Graphical Models
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Zoubin Ghahramani.
Graphical models provide a powerful framework for reasoning about latent explanations of data,
and for integrating different sources of information. But several practical issues result in them
being applied less often than they could be. I will discuss a recent application around inferring the
location of boxes in a supply chain from RFID data. This application is instructive
because it highlights a few of these limitations and presents opportunities for new methodology.
Finally, I will discuss how we are attempting to address another practical barrier to applying
graphical models, namely, scalability of inference techniques. I will describe some very recent work
on new methods for approximate inference that make use of approximate second-order information,
within a quasi-Newton framework.
This talk is part of the Machine Learning @ CUED series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|