Advanced Topics in Graphcal  Models 10-801, Spring 2007 Prof. Eric P. Xing School of Computer Science, Carnegie-Mellon University

Basics

• Schedule and class handouts

• Latex header for scribe note: header.tex
  

• First class: January 17, 2007
• Class lectures: Monday & Wednsday, from 10:30-11:50am
• Location:  WEH 5409
  
• Textbook:
  • There is no required text book.
    
  • Required readings (papers and book chapters) will be posted every week

People

      Instructors:

• Eric Xing, Wean Hall 4127, x8-2559, Office hours: TBA
  

 

Class Assistant:

• Monica Hopes, Wean Hall 4616, x8-5527

 

Teaching Assistant:

• TBA
  

Course Description

Probabilistic graphical models provide a succinct formalism with which to describe structured probabilistic models, help unify our understanding of many computational algorithms, and facilitate developing new algorithms and generalizing classical ones to new applications.

This course departs from the basic formalisms/algorithms covered in 10708, and provides an introduction to advanced statistical and computational methods for the modeling and reasoning of complex, multivariate data in large-scale real-world problems. These methods are based on graphical models and other modern methodologies built on Bayesian formalisms, information theory, and optimization techniques. The focus will be on hierarchical Bayesian models, model selection, optimal-margin learning, approximate inference, Bayesian nonparametric methods. We will discuss both the development of theoretical concepts to support such methods and applications of these methods in a number of domains such as information retrieval, NLP, and genetics.

Course requirements

There are three course requirements (there will be no exams):

• Do the reading and actively participate in discussion, and  SCRIBE NOTES for 2-3 classes.
  
• Develop a piece of original research or write a 5-10 page review of a relevant subject.

Page links

• The Fall 2005 Graphical Model Web Page
• The Fall 2006 Graphical Model Web Page