CS 7343. Machine Learning.

• Course title: CS 7343, Machine Learning
• Instructor: Prof. Ashwin Ram, College of Computing
• Quarter: Usually Winter; check with CoC
• Credit hours: 3-0-3
• Prerequisites: Introductory AI course or permission of instructor
• Description: Fundamental issues in Machine Learning, including the algorithmic, psychological, philosophical and methodological foundations of the field. Topics include empirical, inductive, or similarity-based learning; concept learning; discovery; analytical, theory-based, or explanation-based learning; analogy, knowledge-based analogy, case-based learning; subsymbolic approaches, connectionism, genetic algorithms; and learnability theory. The intent of the course is to provide a strong foundation in machine learning, experience in thinking creatively about issues in this area, and the theoretical and methodological background for further research.
• Audience: The course is appropriate for anyone interested in learning as a research area, and also for anyone interested in applying learning techniques to other research or application areas. If you know absolutely nothing about AI (specifically, if the terms "knowledge representation" and "search" don't ring any bells), you probably want to take CS6361 before taking this course, but if you've had an AI course before and/or some experience with an AI project, you'll be ok. If you're unsure about whether this course is right for you, talk to the instructor.
• Structure: The text (Readings in Machine Learning) is a collection of "classic" papers in the field. We will read and discuss a series of papers from this book (including most of the machine learning papers from the AI depth exam reading list). Each major topic will be introduced with a lecture or two, followed by discussion-based classes in which we will read and discuss two related papers per class. Each paper will have two students assigned to it, a presenter and a responder. The presenter will spend about 15 minutes presenting the paper, highlighting the main ideas and claims from the point of view of the author, discussing the technical details, and relating it to other papers discussed in the course. The responder's role is that of discussant; he or she will respond to (support or argue against) the paper (and, optionally, the presenter's presentation) for about 10 minutes. This will be followed by 15-20 minutes of general class discussion, including "mini-lectures" by the instructor as appropriate. Assignments and term projects will focus on implementation and evaluation of machine learning theories. There will be no exams.
• For more info: Contact Ashwin Ram, ashwin.ram@cc.gatech.edu, 853-9372, or swing by my office (Room 114 CoC).