CMPU-395: Machine Learning

The brain is arguably the most capable information processing device known and has been the source of inspiration for a number of algorithms. In this course we review models of brain function, from numerical simulations of activity in real neurons and small neural networks, via abstract artificial neural networks to machine learning algorithms for classification and prediction based on experience rather than predefined rules. Topics covered include Hodgkin-Huxley and integrate-and-fire models, central pattern generators, attractor models of associative memory, layered networks and error driven learning, competitive learning and formation of topology preserving maps, and behavior formation via reinforcement learning.

Örjan Ekeberg is a STINT Fellow at Vassar during the fall 2009. He is an associate professor in Computer Science at the Royal Institute of Technology in Stockholm, Sweden, where he also works for the Stockholm Brain Institute.

• Office: OLB 106
• Office hours: Mon 15:30–16:30, Thu 11:00–12:00 or by appointment
• E_mail: ekeberg@cs.vassar.edu

Stephen Marsland: Machine Learning, an Algorithmic Perspective; CRC Press, 2009

Lectures are on Mondays and Wednesdays, 1:30 to 2:45, in OH 201.

• Preliminary plan for the lectures.

Six assignments will be given during the course.

• Instructions for the assignments.

Participation during lectures and completion of six bi-weekly assignments are required.

Grading will be weighted like this:

Academic accommodations are available for students with disabilities who are registered with the Office of Disability and Support Services. Students in need of disability accommodations should schedule an appointment with me early in the semester to discuss any accommodations for this course which have been approved by the Office of Disability and Support Services, as indicated in your DSS accommodation letter.