LECTURE 27: DECISION TREES

• Objectives:
  
  
  • Why do we need a smart algorithm to reduce the number of parameters? On what type of information should this smart algorithm operate?
    
    
  • Basic concepts of classification and regression trees (CART)
    
    
  • How do we apply them to acoustic modeling? What are the benefits?

This lecture combines material from the course textbook:

X. Huang, A. Acero, and H.W. Hon, Spoken Language Processing - A Guide to Theory, Algorithm, and System Development, Prentice Hall, Upper Saddle River, New Jersey, USA, ISBN: 0-13-022616-5, 2001.

and these MS project presentations:

• J. Ngan, "Information Theory Based Decision Trees for Data Classification," Master of Science Special Project Presentation, December 10, 1998 (available at Ngan: MS project presentation)
  
  
• A. Le, "Bayesian Decision Tree for Classification of Nonlinear Signal Processing Problems," Master of Science Special Project Presentation, November 12, 1998 (available at Le: MS project presentation)