Vemuri, |
3 |
Machine Learning and Discovery |
R |
TR 12:10p-1:30p |
146 Robbins?? |
25 |
ECS 271, Machine Learning ; Fall 2005
CRN: ?????
Instructor: Prof.
Lecture Times: 12.10 – 1.30 PM - Tu-Th
Office Hours: (in Davis) By appointment
Office Location: 236
Lecture Hall: ???
Prerequisites
1.) Graduate standing in the College of Engineering or permission of
instructor
2.) A First course in probability and statistics such as Stat
131 A.
3.) A background in AI (ECS 170) will make this course easier, but such
a background is not essential. Students who took ECS 170 or an equivalent will
have a decided advantage as some of the topics will be repeated here, at a
faster pace and at a greater depth.
4.) As graduate students you are expected to be good at programming skills (C,
C++ or Java, LISP,
Prolog)
Please talk to me if you have any concerns.
Text Book
Tom M. Mitchell, Machine Learning, McGraw-Hill
ISBN. 0-07-042807-7
Grading
40% for a Project , 60% for Homework and Exams (sample exam )
There will be several homework assignments (approx. one set per week), one midterm and one final.
Project: 40% (Due on the last day of classes)
Midterm: 30%
Final: 10% (Take home. Here, you read and grade two project reports prepared by two other students and turn in your evaluations in 24 hours)
Homework sets: 20%
Your Grades
Course Description .
The field of Machine Learning is concerned with the issue of constructing computer programs that automatically improve with experience. Machine learning draws on concepts from many fields, including statistics, artificial intelligence, cognitive theory, computational complexity and control theory. The goal of this course is to present key algorithms and theory that form the core of machine learning with a balanced presentation of both theory and practice.
A combination of analytical skills and programming skills is expected of the
students wishing to enroll in this class. The project is essentially an
implementation of one or two algorithms discussed in the class.
TOPIC OUTLINE
The topics for the last two weeks will be decided based on what a majority
of the students would like to see covered
The postscript files are viewgraphs supplied by the author
of the text book
1. Introduction, Read Chapter 1,
slides (pdf)
Lecture 0 Slides
Organization of
the course
Term paper,
homework and Exam
Well-posed
learning systems
Deciding to play
tennis or not – A model problem
Perspectives and issues in machine learning
2. Decision Tree Learning, Read
Chapter 3, slides
(pdf)
Lecture 3 Slides
Decision Trees – Overview
Decision Trees - Construction
Decision Trees - Pruning
Homework#2 Do problems on Decision Tables. Build a decision tree, prune it and generate rules out of it.
Solution to HW2 problem, Part 1: Decision tree construction part
Solution to HW2 Problem,
Part 2: Decision tree pruning part
3. Artificial Neural Networks, Chapter 4,
Lecture 4 Slides
Perceptron Learning
Learning via Multi-layer feed forward networks
Perceptron Learning and Support Vector Machines
Homework#3 Problems on activation functions (pdf file). Posted on 16th April, Due on 22 April 2004
Homework#4 Problem on back propagation and couple of theory problems. Posted on 19th April, Due on 29 April 2004
4. Computational Learning Theory
PAC-learnability
Intro to
VC Dimension
Homework#5 Problems on
PAC-learning and VC-dimension. Posted on 30th April, Due on 6 May 2004
Hw#3. Do a probelm on Hypotheses
spaces and PAC learning.
Solution to HW#3
hw4.htm Do Perceptron problem, Shattering Exercises and BP
Solution to Hw4 – shattering question
only
5. Bayesian Learning, Chapter 6,
Bayesian Belief Networks
Solution to Mid-term Examination- Spring
2004
Homework#6 Problems on Probability and
Bayesian nets. Due on 27 May 2004
6. Evolutionary Learning, Chapter 9
Genetic Algorithms
– short tutorial
Genetic
Programming – short tutorial
Homework#7 Problems
on Probability and Genetic Algorithms
4. Concept Learning, Read Chapter 2, slides (pdf)
Lecture 1 Slides – An Overview of Learning
Problems
Lecture 2 Slides
Concept learning as search
Version spaces
Inductive bias
Workedout example on Version spaces
8. Instance-Based Learning , Chapter 8,
Week 9
k-nearest neighbor
learning
Radial basis functions
9. Reinforcement learning
Useful Sites
Department of Computer Science
UC Davis
Last updated on 25March 2003