CSE 590NC: Neural
Computation
Welcome to CSE
590NC (seminar on Neural Computation)! The plan is to discuss one or more
papers per week, diverging as necessary into detail on topics that are
of particular interest to seminar attendees.
We meet on Tuesdays from 1:30 to 2:50 PM in
MGH 278.
What is the seminar all about?
This quarter we will concentrate on topics in sensorimotor learning and control.
Specifically we'll focus on robotics, vision, and statistical methods for learning by
imitation, as well as how to close the "perception-action loop" between agents engaged in
imitation tasks and their environments.
There is a list of presentations with links to appropriate
papers and background reading.
A preliminary list of the broad subject areas that we could pick papers from:
- Theories of shared attention
- Learning forward and inverse models
- Models of motor/premotor cortex
- Learning motor models with and without proprioceptive feedback
- Computational properties of mirror neurons
Papers from other related subject areas may also be picked if there is sufficient interest in the class.
Mailing List: Be sure to join the class mailing list
and browse the email
archive. To join, send a message to majordomo@cs.washington.edu,
and include the following in the body of your message: subscribe
cse590nc
You should get a response quickly that you have been
added.
Online resources: Tutorials, on-line conference proceedings (NIPS) and journals.
Credit Hours: 1
Grading: Credit/No Credit (enrolled students are expected to present one or more papers and participate in class discussions)
Previous quarters of 590NC: Autumn 2001 (Topics in neural networks) Winter 2002 (Statistical frameworks for state estimation)