Machine Learning 2003:
Honours Course


Scope of course: The field of machine learning is concerned with the question of how to construct computer programs that automatically improve with experience. In recent years, many successful machine learning applications have been developed, ranging from data-mining programs that learn to detect fraudulent credit card transactions, to information-filtering programs that learn users' reading preferences, to autonomous vehicles that learn to drive on public highways. At the same time, there have been important advances in the theory and algorithms that form the foundations of this field. The goal of this course is to present some key algorithms and theory that form the core of the AI toolbox. Machine learning draws on concepts and results from many fields, including statistics, artificial intelligence, philosophy, information theory, biology, cognitive science, computational complexity, and control theory.

Textbook: T. Mitchell, Machine Learning, McGraw-Hill, 1997. Website.

A small number of copies of this book are still available for R360 at The Little Big Bookstore, Bellville Business Park A8, c/o Mike Pienaar and Voortrekker Road, Bellville. Apparently the books are not on their main list of computing books, but if you ask for Vivian she will be able to help you.

Evaluation and syllabus (PDF version)


Assignments:

Assignment 1 (PDF version)
Assignment 2 (PDF version)
Assignment 3 (PDF version)
Assignment 4 (PDF version)


Resources:

LaTeX guide: A not very short introduction to LaTeX (Note: you can use xdvi to view dvi files.)

BibTeX guide: BIBTeXing (BibTeX is for bibliographies: you won't use it for this course, but if you become a LaTeX user you will find it useful later.)

Example report: ps version pdf version

NB: This report is provided as an example of what a good report for Assignment 1 might have looked like. For the purpose of illustration I have made up plausible experimental results: you are NOT allowed to do this! Making up fake experimental results would be academic dishonesty and consequences will be severe if you are caught doing this. There are ways to spot fake results!

You will notice that the experiments in the example are not perfect. For example, there is no reason not to use more training games to see if results continue to improve. However, the example is set up to give you an idea of the amount of experimental work that is expected of you in a typical assignment. Similarly, the report might have been written better, but is of a sufficient standard to get a maximum mark.

Lecture slides:
ps version: chapter 1 chapter 2 chapter 3 chapter 4 chapter 5 chapter 6 chapter 7 chapter 9 chapter 13
pdf version: chapter 1 chapter 2 chapter 3 chapter 4 chapter 5 chapter 6 chapter 7 chapter 9 chapter 13

Data:
Character recognition problem (assignment 2): training set test set
Example circle data sets (assignment 3a): training set test set
Octave code for generating and plotting circle data sets (assignment 3a): generate_data.m
Face recognition problem: face_database.tar.Z


Last updated: April 2003
Back to Konrad Scheffler's homepage