CS 281, Spring 1998
Machine Learning
Instructor Stuart Russell
727 Soda Hall,
russell@cs.berkeley.edu,
(510) 642 4964
Office hours Thursday 9.30-12, or russell@cs.
Lecture: Mon, Wed 9.30-11
Location: 310 Soda
Units: 3
Suggested prerequisites: CS188 or equivalent, or permission of
instructor.
Description
The course will attempt to provide a broad introduction to machine learning
from the perspectives of artificial intelligence, theoretical computer
science, and statistics. The treatment will be fairly mathematical in places
but practical applications will be emphasized.
- Learning in intelligent systems (0.5 weeks)
- Learning logical concepts (3 weeks)
- Learning probabilistic models (3.5 weeks)
- Instance-based methods (1 week)
- Neural network models (3 weeks)
- Ensemble methods (1 week)
- Rules and logic programs (1.5 weeks)
- Learning language (1 week)
What will actually happen
The class will meet twice a week to discuss the readings given on the
accompanying reading list.
Prior to each week's discussions, you should prepare a one-page
answer to the discussion questions
for that week (15% of grade). There will be two
simple implementation exercises (20% each) and a term
project (45%) consisting of a substantial project or analytical paper.
Books
Required: Machine Learning, by Tom Mitchell, McGraw Hill, 1997.
Required: Neural Networks and Pattern Recognition, by Chris Bishop,
Oxford, 1995.
Recommended: Artificial Intelligence: A Modern Approach, by
Stuart Russell and Peter Norvig, Prentice Hall, 1995.
Handouts