CSCI 6508 – Neurocomputing: Fundamentals of Computational Neuroscience

Dr. Thomas Trappenberg (tt@cs.dal.ca, 494-3087, Room 313)

Office Hours: Wednesdays/Fridays 9:30-10:30am

News:

Nov 28: Here is the som.m program in case you want to play around with it. Thank you for your participation in class, and good luck with the final test.

Nov 19: Additional exercises. This is not an assignment and will not earn additional points. We will discuss the solutions next Tuesday in class.

Nov 5: The Assignment 6 is a small project with focus on some research and scientific presentation in form of a short paper (guidelines are included). You have two weeks to work on it (it is due on Tuesday, November 19 in class). We can meet the following two Wednesdays to discuss your progress or any other issues that arise.

Oct 24: Assignment 5 is here and due next Thursday (October 31). We can meet next Wednesday to see how far you got.

Oct 16: Here is the rough schedule of topic that I showed you several weeks ago. We are still on track.

Oct 15: The assignment 4 is now online and due on October 22. This assignment has to be solved and submitted individually! Note also that we will have a short test on October 24 in class. There will be no tutorial tomorrow, but please come to see me at my office hours (Wednesday/Fridays 9:30-10:30) if you have questions.

Oct 3: The book is finally available at the Dal bookstore. Please read chapters 3 and 4 for next week (you can omit chapters 3.2 and 4.2) The third assignment (due on October 10) should guide you to the important points on which we will focus. We will have a playtime next Wednesday.

Sept 24: The second assignment is an extension of the previous assignment to multiplayer perceptrons and is due next week on October 1.

Sept 17: Here are the solutions for tutorial 1. The third question of the tutorial is relevant for the first assignment, which is due on Tuesday, September 24. Please come to the `play time’ tomorrow at 5:30 in the unix lab to work on this assignment. Note also my office hour Wednesdays/Fridays 9:30-10:30, beginning this Friday (sorry, not this Wednesday; you can try to catch me other times.)

Sept 10: There will be a tutorial tomorrow (tutorial1) at 5:30 in the unix lab to practice some Matlab, matrix calculus, and to write a program for pattern recognition. You need the pattern1 file.

Outline:

Human cognitive abilities have long guided the development of computer systems. This course introduces the principles of information-processing in the brain, including the functionality of single neurons, networks of neurons, and large-scale neural architectures for specific cognitive functions. We discuss the information representation in the brain (distributed versus localist), information theoretical studies of spiking neurons, synaptic plasticity and adaptive architectures, and various forms of memory. We will also study some specific mental abilities and some research topics (time permitted) such as vision, motor control, navigation, sleep, and consciousness. These issues will be contrasted with approaches that are discussed in the machine learning literature. The course includes a MATLAB tutorial, and the students are encouraged to explore some of the functionalities of basic neural networks.

Proposed syllables and modules:

• Introduction
• MATLAB tutorial (Chapter 12)
• Basic artificial neural networs: multiplayer perceptrons (universal approximators, delta rule, backpropagation) (Chapters 6,10)
• Information processing and models of single neurons (conductance based models, integrate-and-fire neurons, rate models), (Chapters 2-5)
• The neural code (information theory, representations, population coding), (Chapter 5)
• Associatiors (Hebbian plasticity), (Chapter 7)
• Recurrent networks (short-term memory, hippocampus, path-integration) (Chapters 8-9)
• More on mapping networks, reward learning, and motor control (Chapters 6,10)
• System level organizations and specific mental functions (Chapter 11)
• Special project topics and projects (depending on the interest of the students)

(Examples: computer vision, behavioral modeling, data-classification, bottom-up and top-down processing, sleep, …)

Required textbook: (available at Dalhousie bookstore)

Thomas P. Trappenberg Fundamental of Computational Neuroscience Oxford University Press ISBN 0-19-851583-9 Table of contents Programs of the book

Course Evaluation:

COURSE COMPONENT	WEIGHT
ASSIGNMENTS/ PROJECTS	70%
TEST(S)	20%
CLASS PARTICIPATION	10%

Intellectual Honesty:

It is important for students to be aware of the Intellectual Honesty regulations at Dalhousie University. Please see http://www.cs.dal.ca/~grad/gradSite/grad.html#Plagiarism for more information. Any suspected cases of plagiarism will be forwarded to the Senate Discipline Committee.