University of Texas at Arlington
Computer Science Engineering
Fall 2001


CSE 6363 - Machine Learning
Section 501, TuTh 5:30-6:50pm, 315 Nedderman Hall
URL: http://www-cse.uta.edu/~holder/courses/cse6363.html

Description:
A detailed investigation of current machine learning theory and methodologies. Introduces the background and basics of machine learning, including representation, inductive bias and performance evaluation. Analyzes and compares different machine learning methodologies, including statistical, connectionist, symbolic and optimization. Implementations of several methods will be provided for experimentation. Current issues in machine learning research and alternative learning methods will also be examined as they relate to course topics.

Prerequisites:
Artificial Intelligence I (CSE 5360) and II (CSE 5361).

Textbook:
Tom M. Mitchell, Machine Learning, McGraw-Hill, 1997.

Grading:
Six Homeworks (60%), Project (20%), Presentation (10%), Critiques and Class Participation (10%).

Instructor:
Larry Holder, 333 Nedderman Hall, 272-2596, holder@cse.uta.edu.
Office hours: TuTh 3:30-4:30pm or by appointment.

Course Details

The purpose of this course is to expose the student to several machine learning paradigms and provide in-depth understanding of selected methods. Homeworks will provide hands-on experience with selected learning methods and will test the student's understanding of topics discussed in class. All work in this course must be done individually.

The class project serves to further the student's understanding of one or more methods through an experimental or theoretical analysis of the learning methods. Students are required to obtain approval on a project proposal. Ideas for projects will be distributed in class.

Each student will make a class presentation on one or more technical papers in the area of machine learning to be approved by the instructor. The presentation must demonstrate an indepth understanding of the topic, including background material from the paper's references, and provide a critical review. Longer papers tend to be more self-contained; whereas, shorter papers will require more auxiliary material.

Lastly, the student is required to participate in class by submitting brief critiques of selected papers to be covered in class and participating in class discussions. The purpose of the critiques is to stimulate critical thinking and discussions on class topics. A critique is not merely a summary, but your own reactions to the content of the reading. Critiques should be approximately one page of single spaced type.

All assignments must be turned in on time. There will be no credit for late submissions. All work in this class must be done individually.

If you require an accommodation based on disability, please see me during the first week of the semester so we can be sure you are appropriately accommodated.

Schedule

Class Date Topics Readings Assignments Due
1 8/28 Introduction    
2 8/30 Introduction Ch1  
3 9/4 Concept Learning Ch2  
4 9/6 Concept Learning    
5 9/11 Decision Tree Learning Ch3 HW1
6 9/13 Decision Tree Learning    
7 9/18 Neural Networks Ch4  
8 9/20 Neural Networks    
9 9/25 Evaluating Hypotheses Ch5 HW2
10 9/27 Evaluating Hypotheses    
11 10/2 Bayesian Learning Ch6  
12 10/4 Bayesian Learning    
13 10/9 Bayesian Learning   HW3
14 10/11 Learning Theory Ch7  
15 10/16 Learning Theory    
16 10/18 Instance-Based Learning Ch8  
17 10/23 Genetic Algorithms Ch9 HW4
18 10/25 Learning Rule Sets Ch10  
19 10/30 Learning Rule Sets    
20 11/1 Analytical Learning Ch11  
21 11/6 Combined Learning Ch12 HW5
22 11/8 Combined Learning   Project Proposal
23 11/13 Reinforcement Learning Ch13  
24 11/15 Student Presentations papers*  
25 11/20 Student Presentations papers* HW6
26 11/27 Student Presentations papers*  
27 11/29 Student Presentations papers*  
28 12/4 Current Issues papers*  
29 12/6 Conclusions    
  12/11     Project

* Indicates that a written critique is due on that day covering papers to be distributed in class.