CSE 546 - Data Mining

University of Washington Department of Computer Science & Engineering

CSE 546 - Data Mining - Autumn 2003

CSE Home

About Us

Contact Info

Instructor: pedrod at cs.washington.edu
Office: Allen 648
Office hours: Wednesdays 2:00-2:50 and by appointment TA: mattr at cs.washington.edu
Office: Allen 220
Office hours: Mondays 4:30-5:20 and by appointment

Class meets:
Mondays and Wednesdays from 3:00 to 4:20 in MEB 242

Readings

Week 1: Chapter 1 of Hand and Behind-the-scenes data mining
Week 2: Chapter 3 of Mitchell
Week 3: Chapter 10 of Mitchell; review first-order logic
Week 4: Mining high-speed data streams, Mining complex models from arbitrarily large databases in constant time
Week 6: Chapter 6 of Mitchell; review probability and statistics
Week 7: Chapter 4 of Mitchell; review calculus
Week 8: Section 2 of Machine-learning research: Four current directions
Week 9: Chapter 7 of Mitchell, A unified bias-variance decomposition, A tutorial on support vector machines
Week 10: Chapter 9 of Hand

Lecture Notes

Week 1: Introduction, inductive learning
Week 2: Decision trees
Week 3: Rule induction
Week 4: Scalability
Week 5: Instance-based learning
Week 6: Bayesian learning
Week 7: Neural networks
Week 8: Model ensembles
Week 9: Learning theory and SVMs
Week 10: Clustering

Topics

The topics covered will have a non-null intersection with the following list:

The data mining process
Decision trees
Rule induction
Instance-based learning
Bayesian learning
Neural networks
Genetic algorithms
Model ensembles
Learning theory
Support vector machines
Clustering
Association rules
Data warehousing and OLAP
Web mining
Scalability

Project

Class evaluation will be by means of a project. Projects can be proposed by the students - for example, applying data mining techniques to your area of interest - or chosen from this list:

Mining multi-relational databases
Extracting knowledge bases from the World Wide Web
Mining social networks from the Internet
Incorporating prior knowledge into data mining algorithms
Learning models of an adversary's behavior
Mining sequences of program versions to automate debugging
Merging and refining databases for data mining
Predicting the evolution of scientific communities
Simultaneously classifying sets of related entities
Discovering unobserved relations from observable ones
Learning across different subpopulations
Learning given a query distribution
Modeling the hidden Web through querying
Modeling Web surfing behavior
Scaling up learning algorithms via sampling and compression

Projects can be carried out in groups of two or individually; we encourage working in groups. In addition to a written report, students will give a short oral presentation of their work.

Schedule of project presentations

Textbooks

Tom Mitchell, Machine Learning, McGraw-Hill, 1997 (required).
David Hand, Heikki Mannila and Padhraic Smyth, Principles of Data Mining, MIT Press, 2001 (recommended).

Papers

C. Burges, A tutorial on support vector machines for pattern recognition, Data Mining and Knowledge Discovery, 1998.
T. Dietterich, Machine-learning research: Four current directions, AI Magazine, 1997.
P. Domingos, A unified bias-variance decomposition for zero-one and squared loss, Seventeenth National Conference on Artificial Intelligence, 2000.
P. Domingos and G. Hulten, Mining high-speed data streams, Sixth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2000.
G. Hulten and P. Domingos, Mining complex models from arbitrarily large databases in constant time, Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2002.
G. John, Behind-the-scenes data mining, SIGKDD Explorations, 1999.

Software

VFML is a set of tools developed at UW that you may find useful for your project. VFML is still in beta; if you're planning to use it, please contact ghulten at cs.washington.edu. Pointers to various pieces of data mining software can be found at KDnuggets.

Anonymous Feedback

Comments can be sent to the instructor or TAs using this anonymous feedback form.

Course Mailing List

To subscribe to the course mailing list, visit the mailing list home page. Alternatively, you can use the email interface to subscribe; send email to cse546-request@cs with the word "help" in the subject to receive a list of email command options.

Mailing List Archive

Department of Computer Science & Engineering
University of Washington
Box 352350
Seattle, WA 98195-2350
(206) 543-1695 voice, (206) 543-2969 FAX
[comments to Pedro Domingos]