This page is a resource for exchanging ideas relating to CS 478. Anyone may read the information on this wiki; to add to or edit the pages, you'll have to create an account. Click on the “Login” button below to begin the process.
This course is a new instantiation of CS478. It combines material from the old CS478 and CS476. Going forward, there will only be a single ML/Datamining course at the 400 level and this course will provide students with tools and understanding useful for application in industry or as a basis for further study in a graduate program.
Material that will transition, in some form, from CS476: use of WEKA, feature selection, association rule mining, lin/log regression, clustering, model evaluation and a group project.
Material that will mostly not transition from CS476: SQL and OLAP and the major focus and web and business applications.
Material that will transition, in some form, from (the old) CS478: classification, induction, error/utility measures, decision trees, backpropagation, comparing classifiers, instance-based learning and concepts like generalization and bias.
Material that will mostly not transition from (the old) CS478: some of the focus on learning/intelligence, some of the focus on implementation of models, genetic algorithms, naive Bayes, reinforcement learning.
In addition some of this material will be augmented/supplemented by material that will trickle down from CS678 (e.g. Support Vector Machines, Boosting, No Free Lunch Theorem.
Project groups have now been formed (using a Genetic Algorithm, which worked really well – everyone is on a project they listed as one of their top 2, and all but 5 people are on their most preferred project). Six groups have four people and one has five. The links below are your group wikis for intermediate reports, collaboration, etc. Note the suggested time-line on the course schedule. It is *critical* that you stick to this schedule and make consistent progress on this throughout the remainder of the semester. This is not something that you can do in a week or two at the end. Most importantly, have fun.