CIS 410/510, Machine Learning

CIS410/510, Spring 2007

Machine Learning

[ Assignments | CIS | News]

Course Description: Machine learning is the field which concerns how to construct computer programs that automatically improve with experience. It has successful applications, such as to data mining, information retrieval, robotics and biomedical informatics. The intent of this course is to present a broad introduction to machine learning and its established algorithms, including discussions of each of the major approaches currently being investigated. The topics include neural networks, decision trees, Bayesian learning, support vector machines, inductive logic programming, reinforcement learning and learning theory. One secondary goal is to discuss the connections between machine learning and other fields, such as AI, statistics, databases and Web.
Time and Place: Tuesdays and Thursdays at 12:00pm - 1:20pm in 200 Deschutes Hall.
Instructor: Dejing Dou, 303 Deschutes, phone 346-4572, email dou@cs.uoregon.edu.
GTF: TBA
Office hours:
- Dejing Fridays 2:30-3:30 or by appointment (I like to talk with students).
Directory: The (unix) directory for the class is /cs/classes/www/07S/cis410ml.
News: Keep an eye on the course news for miscellaneous announcements.
Evaluation:
- The homework will be a combination of written and programming work.
- The Course Project will be due in parts: Part 1 (proposal) is due Friday May 18 and Part 2 (final report) on Friday June 15. We assume that you will work on it individually. If you wish to work in a group, please check with the instructor.
The approximate weighting of the work done will be

Homework Assignments 40%

Midterm (May 10) 20%

Final (June 13) 20%

Course Project 20%
Text:
- (Required) Tom Mitchell, Machine Learning, McGraw Hill, 1997.(text website)

Approximate Syllabus and Lecture Slides

Week of	Chapter	Description and Slides	Events

Apr 2	1	intro, learning system design;
Apr 9	2	concept learning, general-to-specific ordering
Apr 16		Prof. Art Farley will give a guest lecture on Apr. 17; classic paper reading:SVM,FOIL	Dejing will be a NSF panel meeting
Apr 23	3	decision tree learning; classic paper reading: C4.5	Dejing will be SDM07 conference on April 26
Apr 30	3,4	bagging and boosting, artificial neural network
May 7	5,6	statistics intro, Bayesian learning	Midterm Thursday, May 10
May 14	additional	Bayes learning(cont'd), EM, support vector machine(SVM)	Course Project Part 1: Fri, May 18
May 21	10	SVM(cont'd), learning sets of rules
May 28	10, 13	ILP, reinforcement learning
Jun 4		course project implemenation, no lectures
Jun 11		finals week;	Final exam Wed 8am-10am, Jun 13; Course Project Part 2: Fri, Jun 15

Late policy: All assignments are due by 11:59pm of the stated due date, though it is much safer to turn them in at class time. Late papers are subject to a 10% per day penalty. Late assignments will not be accepted once the graded papers are returned or once sample solutions are handed out. It is your responsibility to turn in the assignments on time. Make sure that electronically submitted assignments are actually delivered.
Course resources:
- Text website.
- Machine Learning Information Site

Homework Assignments	40%
Midterm (May 10)	20%
Final (June 13)	20%
Course Project	20%