CGSC 5001:
Cognition and Artificial Systems
Course Outline

School: Carleton University
Department: Institute of Cognitive Science
Site: http://www.carleton.ca/ics/courses/cgsci5001/
Term: Fall 2005

Course Description:

"An introduction to the contribution of artificial intelligence and computer modelling of cognitive processes to cognitive science." (From the calendar)

More specifically, this course will give students hands-on working knowledge with various cognitive models and AI techniques. Students will not be writing AI algorithms, but rather they will be exploring how to use them and in what situations various modelling approaches are suitable. We will also explore the standards of proof needed for computational modelling to inform cognitive science.

Prerequisites:

Graduate or 4th year undergraduate standing. For undergraduates, one programming course (or equivalent) is a prerequisite. You will be expected to write and/or modify very short programs (usually 30-40 lines) whose main purpose will be to use existing computational libraries to investigate various models. You will not be writing AI algorithms from scratch; rather, you will be interfacing to existing code and applying it to different situations.

Instructor:

Terry Stewart

tcstewar@connect.carleton.ca
http://terrystewart.ca
2209/2212B Dunton Tower
Office hours by appointment

Classes:

One 3-hour class per week (Tuesdays, 2:30-5:30, 505 Southam).

Course Objectives:

To understand the basic issues of computability and the limitations/advantages of computational models as compared to other techniques (such as mathematical proofs or experimental results)
To get experience using neural networks, genetic algorithms, and symbolic cognitive models in various situations.
To get a feel for the appropriateness of the afore-mentioned models in various situations
To apply the concept of "replicability" to cognitive computational modelling, as current practices sorely neglect this idea

Readings:

Each week, a collection of academic papers are given as recommending reading. Also, the course lecture notes are available online and may be read beforehand. All readings will be available in electronic format. There will be no course textbook.

Software:

In-class examples are in the programming language Python (chosen for its simplicity, readability, and extensive libraries). Students are not expected to have previous experience with the language. We will be using various pre-existing libraries for running the computer models. All software will be freely available online. It is hoped that students will be able to use their home computers, but if not lab facilities will be found on-campus with the required software (although possibly with annoying access hours).

Evaluation:

Assignments 70% - Weekly short assignments investigating the properties of a particular system in various situations. Students will have one week to answer a set of questions about the model's performance. Students will be provided with sample code for reading input data files, interfacing with modules, and displaying results. The answers are due in class one week after they are given. Due to University policy, assignments cannot be submitted by fax or email. Students are responsible for keeping a copy of their own assignments, just in case. There will be 10 assignments, and the last one will be due the last week of class (i.e. there is no assignment given out on the last week). Papers will be returned to students with marks 1 week later. Each assignment is given an equal weighting.
Final Paper 30% - The final paper will involve choosing a published paper in the field (either on your own or from weekly readings) and replicating the findings. You can use any freely available tools, and should investigate the effects of modifying the model in ways not mentioned in the original paper. This paper will be formatted as a 4-6 page (single-spaced) official paper and will discuss any differences between your results and the original results, what required information was left out of the original paper (trust me, there'll be some), and any other problems involved in replicating their work. The paper is due at the end of the exam period.

Group Work

I strongly encourage students to work together on the weekly assignments. Having two (or more!) people playing around with the various approaches can lead to a better understanding of the strengths and weaknesses. That said, I do expect each student to hand in their own answers and their own interpretation of the results that they may have found as a group in the weekly assignments. Write about things that you may have discussed together, but write about them in your own words.

The final paper cannot be done as a group.

Plagiarism

"The Senate of the University has labeled plagiarism an instructional offence. For the University’s purposes to plagiarize is to “use and pass off as one's own idea or product work of another without expressly giving credit to another”. Being caught plagiarizing can result in one being expelled. suspended from all studies at the University, suspended from full-time studies, awarded a reprimand, refused permission to continue or to register in a specific degree program but subject to having met all academic requirements shall be permitted to register and continue in some other program, placed on Academic Warning, or awarded an F or Abs in a course or examination."
- from the official course outline guidelines

Course Calendar:

Week 1: Introduction, Algorithms, Chaos, and the Limits of Mathematics
Week 2: Cellular Automata
Week 3: Genetic Algorithms: Introduction
Week 4: Genetic Algorithms: Applications
Week 5: Evolutionary Strategies, Neutral Networks and Extrema Selection (note that that's Neutral, not Neural)
Week 6: Neural Networks: Introduction (no assignment)
Week 7: Back-propagation Neural Networks
Week 8: Recurrent Networks and Self-Organizing Maps
Week 9: ART and ARTMAP
Week 10: Reinforcement Learning, Whole-System models
Week 11: Expert Systems and ACT-R
Week 12: Putting the "Science" in Cognitive Science (no assignment)
Week 13: Review Week: Other topics of interest, or time to talk about final papers

Students with Disabilities

Students with disabilities requiring accommodations in this course are encouraged to contact a coordinator at the Paul Menton Centre for Students with Disabilities to complete necessary letters of accommodation. After registering with the PMC, please make an appointment to meet and discuss your needs with me. The deadline for submitting completed forms to the PMC is March 7th.

Student Life Services

From http://www.carleton.ca/studentlife/:

The Department of Student Life Services is committed to enhancing the learning that takes place inside and outside the classroom. We believe our services and programs can make a difference in the daily lives of students at Carleton. Therefore, we work in partnership with students in setting and striving for their academic, personal, and career goals.
A wide range of programs and services are offered to assist students in their adjustment to academic life; in improving their learning skills; and in making decisions with regard to academic and career concerns.

Writing Tutorial Service

From http://www.carleton.ca/wts/:

At the Writing Tutorial Service, we help you write better papers. In our tutoring sessions, we will work with you to help you develop your writing process. We will also teach you strategies that will help you write the way you want to write. WTS tutors are graduate students trained to assist you at any stage in the writing process.
To make an appointment for a tutoring session phone 520-6632, or go to Room 215, Paterson Hall.

Menu

CGSC 5001: Cognition and Artificial Systems Course Outline