Empirical Methods in Machine Learning & Data Mining
CS578
Computer Science Department
Cornell University
Fall 2005

General Information    Lecture Notes    ML Links    Assignments    Project

Announcements:

• By popular demand, we will hold a pizza session to go over results for the final project. We meet today (Thu) in Upson 5126 at 5:30. Pizza will be served. I'll begin the summary of the final projects at 5:45 so that people with exams until 5:30 can get there. You also can pick up the final projects at the meeting today.
• Results for the final project are now available at:  http://kodiak.cs.cornell.edu/cgi-bin/cs578/projectresultstable.pl
• Final Project Extension: predictions may be submitted as late as 1AM Thu Dec 8 (Wed night 1 hour after midnight).  The Final report is still due Thu at noon for full credit.  There is a penalty of 1 point if the report is handed in after noon Thu but by 4pm Fri Dec 9, and a penalty of 2 points if the report is handed in by 10am Monday Dec 12.  No reports will be accepted after 10am Mon.
• Rich's office hours this week are Monday Dec 5 at 10:30-11:30 and 2:30-3:00. Alex is away all week. Lars and Cristi will hold office hours as usual.
• The review session for the midterm will be Fri Nov 2 5:30-6:30 in Olin Hall 165 (same room as the course).  Note that this session is not the right place to ask about how your midterm was graded.  If you have specific questions about grading you should visit one of the TAs next week.
• Online course evaluations are due by Dec 3.  Students who complete the online evaluation will receive +1 points in their grade for the course.  Although we are told which students do evaluations, we are never given identifying information with the evaluations, so feel free to give us honest feedback about the course.  Thanks!
• If you are looking for partners for the final project email us and we'll make an announcement in class on Tue.
• Final project prediction submissions: 
•     to register:  http://kodiak.cs.cornell.edu/kddcup/cs578/register.html
•     to submit predictions: http://kodiak.cs.cornell.edu/kddcup/cs578/submission.html
• Final report is due at noon on Thu Dec 8.  Reports will not be accepted after Thu Dec 8.
• Turn in final report to Amy Fish in Upson 4115.
• Final Project now out.  Predictions officially due Sat Dec 3. Submission website will close midnight Tue Dec 6.  project.tar.gz

Homework 3 is now available.  It's due at the beginning of class on Tue Nov 22.

Clarification for midterm: XOR of three inputs is defined to be +1 if the number of +1 inputs is odd, and -1 if the number of +1 inputs is even.

Small error in the midterm: on the decision tree question where you are asked to calculate Info_gain, Gain_Ratio, and Gini_Score,  Gini_Score is a typo and you should be calculating RMS_Score instead.  This should be obvious from the rest of the question, but just in case.

The midterm handed out in class is missing the last page.  Here is the complete text file: trick-or-treat.2005.txt  Very sorry for the confusion.  The midterm is due 2:55pm on Thu Nov 3.  Unlike homework, we will not accept late midterms.  Good luck.  I hope you enjoy it.

Homework Assignment 2 is now available.  It's due at the beginning of class on Thu Oct 20.

There are problems getting IND to compile using the new compiler in recent versions of CYGWIN.  You can avoid these problems by telling CYGWIN to use compiler 3.3.3 (instead of the newer 3.4.4) in the CYGWIN Setup Tool.  Note that you also should use the Setup Tool to install tcsh, bison, and make5

The Unix/UnixSTAT/Scripting tutorial will be held Tue 7:30-9:30 in Olin Hall 165.  The tutorial is optional and is intended to help people who are not familiar with Unix or writing scripts get up to speed.

Homework Assignment 1 is now available.  It's due at the beginning of class on Tue Sep 20.

We requested that the Mitchell textbook be put on reserve, but it may take a few days for it to be available because the library's copy from last year was not returned.

Time and Place

Personnel

Go to top

General Information

Course Description:
This implementation-oriented course presents a broad introduction to current algorithms and approaches in machine learning, knowledge discovery, and data mining and their application to real-world learning and decision-making tasks. The course also will cover empirical methods for comparing learning algorithms, for understanding and explaining their differences, for exploring the conditions under which each is most appropriate, and for figuring out how to get the best possible performance out of them on real problems.

Tentative Course Syllabus

Textbooks:
Machine Learning by Tom Mitchell

Optional references:
The Elements of Statistical Learning: Data Mining, Inference, and Prediction by T. Hastie, R. Tibshirani, J. Friedman.
Pattern Classification 2nd edition by Richard Duda, Peter Hart, & David Stork

Grading policies:

• 25% Midterm (take home)
• 25% Final (open book in class)
• 30% Assignments (individual)
• 20% Final Project (group project comparing various learning methods)
• Bonus points for class participation
• Homeworks, the take-home mid-term, and the final exam must be your own work. For homework, it is OK to talk with other students about the assignment, ask each other questions, and in general learn from each other. But the homework you hand in must be your own work. If other students gave you significant help with your homework you should briefly acknowledge them in what you hand in.

Academic integrity policy

Go to top

Lecture Notes

Unsupervised Learning Clustering slides (cs578_clustering_lecture.4up.pdf)

MTL slides (cs578.mtl.lecture.4up.pdf)

SVM slides (slides_sigir03_tutorial-modified.v3.pdf)

Bagging & Boosting Slides (CS578.bagging.boosting.lecture.pdf)

Special Topics: Missing Values & Feature Selection Slides (missing_featsel_lecture.pdf)

KNN Slides (CS578_knn_lecture.4up.pdf)

Revised Performance Measures Slides (last update 10/18/05) (performance_measures.4up.pdf)

History File from Unix Tutorial (cu.578.05.unix.history.txt)

Revised Decision Tree Slides (last updated 9/15/05) (CS578.05_DT_lecture.2up.pdf)

Introduction to COMS 578 and a Brief History of Statistics, Machine Learning, and Data Mining (CS578.05_INTRO_lecture.pdf)

Go to top

Assignments

HW1 Decision Tree Assignment (due start of class Tue Sep 20): cs578.hw1.tar
IND download for MacOS 10.3: ind.macos10.3.tar
UnixStat download for MacOS 10.3: unixstat.macos10.3.tar

HW2 Neural Nets Assignment (due start of class Thu Oct 20 ): hw2.tar.gz
Perf code for calculating ROC performances: http://kodiak.cs.cornell.edu/kddcup/software.html 

HW3 KNN Assignment (due start of class Tue Nov 22): hw3.ps  hw3.data.gz

Go to top

Final Project

• Final project prediction submissions: 
•     to register:  http://kodiak.cs.cornell.edu/kddcup/cs578/register.html
•     to submit predictions: http://kodiak.cs.cornell.edu/kddcup/cs578/submission.html
• Final report officially is due Sat Dec 3.  Reports will not be accepted after 12 (noon) Wed Dec 7.
• Turn in final report to Amy Fish in Upson 4146.
• Final Project now out.  Predictions officially due Sat Dec 3. Submission website will close midnight Tue Dec 6.  project.tar.gz
• Here's the PERF software for HW2 and the Final Project:  http://kodiak.cs.cornell.edu/kddcup/software.html

Go to top

ML Links

• The UC-Irvine ML Dataset Archive | The UC-Irvine KDD archive | more datasets
• ML Projects
  • The WEKA Machine Learning Project (code for many ML algo's, as well as some datasets)
• Journals
  • Machine Learning | Journal of ML Research | Data Mining and Knowledge Discovery | Journal of AI Research | AI Magazine | IEEE Neural Networks Council (several journals are connected to this page)
• Knowledge Discovery in Databases

• Other University ML classes
  • Wisc Madison More External AI References
• Pointers to ML Courses
• Machine Learning Benchmarking
• AI Bibliography Server | Neural Networks Bibliography Server (Austrian AI Institute)
• AI Resources (Canadian NRC Server)
• Aha's ML Links
• Stuart Russell's:AI on the Web (loads of links)
• Reinforcement Learning Repository

Go to top