Topic:
Machine Learning Basics - II
Abstract:
In this tutorial a review of the graphical
models and associated inference and learning algorithms would be done. While
probability theory offers a natural way to encode and reason under uncertainty,
graphical models provide a concise qualitative description of the joint
distribution over a large number of random variables. This tutorial would
emphasize on two important operations commonly performed on graphical models –
inferring a distribution over a subset of variables given some evidence and
learning parameters associated with the joint distribution. Also some of
the widely used inference algorithms including iterative conditional modes,
belief propagation and variational methods, and approaches to learning models
from data (parameter estimation) would be taken up. Further an
illustration of applications in computer vision and/or computational biology
would also be touched.
Bio of the Speaker:
Anita Kannan
is presently an Associate Researcher in MSR, Cambridge. She did her PhD at University of Toronto where she was
a member of Brendan Frey’s Probabilistic and Statistical Inference group. Her research focuses on machine learning,
with emphasis on probabilistic models. Her current interests are in developing
efficient probabilistic models for solving problems in computer vision and
computational biology.
Homepage:
http://research.microsoft.com/~ankannan/
E-mail: ankannan@microsoft.com
Additional Material (References, Slides & Lecture Notes):