Machine Learning (EECS 349)

Fall 2010
Electrical Engineering and Computer Science Department
Northwestern University

Class Meets: 1:00PM-1:50PM MWF, Tech LR4

Instructor: Doug Downey
Office Hours: 2:00-3:00PM Wednesday (or by appt), Ford 3-345
Email: ddowney <at> eecs <dot> northwestern <dot> edu

Teaching Assistants: Zhiyao Duan and Arefin Huq
Office Hours: (3:00PM-4:00PM Wed., Ford 3-230) and (2:00-3:00PM Mon., Ford 3-204)
Email: zhiyaoduan00 <at> gmail <dot> com, arefinhuq2013 <at> u <dot> northwestern <dot> edu

Completed Course Projects

Policies

Four homework assignments make up 50 points of the grade, a course project makes up 55 additional points, for a total of 105 points. Grades are assigned using the standard scale (given in the "introduction" lecture notes), so 93-105 points is an A, 90-93 points is an A-, etc.

Homework

Submit your homework via email to both zhiyaoduan00 <at> gmail <dot> com and arefinhuq2013 <at> u <dot> northwestern <dot> edu. Put EECS349-PS<problem set number>-<first name>-<LastName> on the subject of your email and attach a compressed ZIP file with the solution. The ZIP file naming convention is: PS<problem set number>-<first name>-<LastName>.zip. For example, if your name is James Bond and you are submitting your solution to Problem Set 1, you will send the TA an email with EECS349-PS1-James-Bond as the subject and you will attach the file PS1-James-Bond.zip which contains all the files that comprise your solution to Problem Set 1. Details on the specific files to include are given in each homework assignment.

Late assignments are penalized by 5% a day, and will NOT BE ACCEPTED more than one week after the original deadline.
Problem Set 1Due 11:59PM Tuesday, Oct 510 pts
Problem Set 2Due 11:59PM Thursday, Oct 2115 pts
Problem Set 3Due 11:59PM Thursday, Nov 415 pts
Problem Set 4Due 11:59PM Tuesday, Nov 2310 pts

Course Projects

Course projects will be performed in groups of students; any number of students can work together. Your assignment is to formulate an interesting task for which machine learning can be used, gather training and test data for the task, and then evaluate one or more machine learning algorithms on the data. The Web pages for the completed course projects from 2008 and other years are available for perusal.

Deadlines:
Proposal (1 pg)Due 11:59PM Tuesday, Oct 2610 pts
Status Report (2 pg)Due 11:59PM Tuesday, Nov 1610 pts
Project Poster/Demo 9AM Thursday, Dec 920 pts
Project Web page (details in poster link above)9AM Thursday, Dec 915 pts

Reading

Week of September 20 Wired data-mining article
Recommended: Ch. 1 & 2 of Mitchell
Week of September 27 Recommended: Ch. 3 of Mitchell
Week of October 4 Recommended: Ch. 8 of Mitchell
Week of October 11 Recommended: Ch. 9 of Mitchell
Week of October 18 Recommended: Clustering Tutorial
Bagging, Boosting, and C4.5
Week of October 25 Recommended: Ch. 6 of Mitchell
Week of November 1 Recommended: Andrew Moore tutorial on Bayes Nets
Week of November 8 Recommended: Ch. 4,5, and 7 of Mitchell
Week of November 22 Recommended: SVM Tutorial

Lectures and Schedule (subject to change)

Week of September 20Introduction
Decision Trees
Week of September 27M-W: Decision Trees (cont.)
F: Instance-based Learning
Week of October 4M: Distance Measures
W: Project Guidelines and Suggestions
F: Greedy Local Search
Week of October 11M: Genetic Algorithms
W: Genetic Programming (Forrest Stonedahl guest lecture)
F: Machine Learning in Industry (Mykell Miller guest lecture)
Week of October 18M-W: Clustering
F: Ensemble Methods (Zhiyao Duan guest lecture)
Week of October 25M: Basics of Probability for Machine Learning
W: Statistical Estimation
F: Naive Bayes Classifiers
Week of Nov 1M: Bayes Nets
W-F: Hidden Markov Models
Week of Nov 8M: Computational Learning Theory and Evaluating Hypotheses
W: Active Learning
F: Neural Networks
Week of Nov 15M: Neural Networks (cont.)
W: Feature Selection (Arefin Huq guest lecture)
F: Project Status Reports
Week of Nov 22M: Support Vector Machines
W: Web Information Extraction
F: No class (Thanksgiving break)
Week of Nov 29M-W: Reinforcement Learning
F: Final project issues