Old Web Page for
CPSC 689-601 Intelligent Neural Systems:
Spring 2003

Syllabus

NEWS:
  1. Final grades have been posted (5/6/03)
  2. Term project postponed to 5/2 (see Final exam schedule).
  3. Regular lecture on 4/22.
  4. Homework #5 will be available on Thursday 3/28 (deadline will be postponed accordingly).
  5. Sejong's paper for Thursday was changed to Bell and Sejnowski (1997).

  6. Resources links updated: neural networks freeware/shareware page.
  7. Project/Term Paper proposal due is 3/20 (Thu). See the board for more detail.
  8. Term Project description is now online (2/26).
[Read-Only Bulletin Board]
*. All email submissions should be in PLAIN ASCII TEXT.
*. Do not send word docs.
Keep an eye on this news box for the latest.

[ Resources | Calendar | Weekly Sched | Lecture Notes | Readling List | Assignments ]

| Intro | Goals | Textbook | Admin |

Instructor:

Dr. Yoonsuck Choe
Email: choe(a)tamu.edu
Office: HRBB 322B
Phone: 845-5466
Office hours: T/TH 1:00-2:00PM, and at other times by appointment only.

Prerequisite/Restrictions:

A little knowledge in linear algebra and probability theory and graduate standing, or by consent of the instructor.

Lectures:

T/TH 11:10am-12:25pm, HRBB 126

Introduction:


How does the brain generate intelligent (or complex) behavior? The focus of this course is to address this very question from various different perspectives. Select topics from computational vision, computational neuroscience, and cognitive science will be reviewed and critiqued. In the first few weeks, basic computational and mathematical preliminaries, as well as neuroscience basics will be covered. Afterwards, a selected collection of current research papers will be discussed. The course is designed to be open-ended to some degree, and a large portion of the time will be dedicated to discussion of the topics.

Goal:

The goal of this course is to
  1. learn basic computational and mathematical tools for investigating the nervous system;
  2. get acquaninted with diverse computational approaches to the understanding of brain function; and
  3. explore how the seemingly disjoint topics can be integrated in a unique synthesis.

Textbook:

Administrative Trivia:

  1. Computer accounts: if you do not have a unix account, ask for one on the CS web page.
  2. We will use Matlab(tm) (there is also an excellent open source clone called GNU/Octave). Matlab is installed on all SunOS machines (and also on the Windoes machines -- I've got to check).
| Topics | Grading | Acad. Policy |

Topics to be covered:

See the Calendar section below for reading and other assignments for each week. The following topics will be covered:
  1. nervous system overview (Ballard chapter 1)
  2. computational tools and basic theoretical concepts (Ballard chapters 2, 4, 8, and 9),
  3. natural sensory statistics and neural coding,
  4. self-organization in neural networks,
  5. issues arising from the encoding-decoding approach,
  6. complementary approaches (active representations and analogy), and
  7. applications in embodied agents.

Grading:

  1. Assignments (30%): about 3 short programming assignments (10% each).
  2. Paper presentation (15%): each student will study and present a paper from the reading list. The term project may be loosely based on this paper.
  3. Paper comments (15%): for the reading assignments each week, a brief (one paragraph) comment/critique must be submitted. Occasionally, the instructor will ask a specific question or ask the student to comment on a particular aspect of the paper.
  4. Term project (40%): 6-7 page term paper describing the project, and project demo and a presentation (30 minutes + 10 minutes Q/A). The project can either be done individually or as a team of two.
Grading will be on the absolute scale. The cutoff for an `A' will be at most 90% of total score, 80% for a `B', 70% for a `C', and 60% for a `D'. However, these cutoffs might be lowered at the end of the semester to accomodate the actual distribution of grades.

Academic Policy:

All incidents of academic dishonesty will be dealt with according to the university policy. No exceptions.
  • All work should be done individually, unless otherwise noticed. Discussions are encouraged, but the submitted material must be in your own words.
  • All references must be properly cited, including internet web pages (URL must be provided). If plagarism is detected, i.e. without proper citation and quotation, you will automatically receive an F. When in doubt, please ask the instructor if it is reasonable to include other's work in your assignments.

Resources:

  1. Neural Networks: Sharewares and Freewares (Thanks to Subru)
  2. Thalamus slices
  3. Neuroscience Tutorial at The Washington University School of Medicine. (Thanks to Barani)
  4. Research resources page
  5. Ganeral reading list (u: p: ): includes short blurb about how to find, read, and critique others' work. This list is not the course reading list.

Calendar:

See: TAMU's official academic calendar.

Weekly Schedule and Class Notes
Week Topic Tue Thu Comments/Homeworks
Week 1 (1/13-) Course Overview
  • First class 1/14 (Tue)
  • Course overview
1/16 (Thu): last day for dropping courses.
  • Introduction to the nervous system (Ballard Chapter 1)
  • Supplementary reading: Charles F. Stevens. The neuron. (From the handout.)
1/17 (Fri): last day to add courses.
Week 2 (1/20-): Bayes' Rule and Information Theory
  • Reading list overview.
  • Some more overview of the course topics.
  • Bayes' rule, probability, information theory, etc. (Ballard Chapter 2)
  • Reading: Bell (1999).
  • HW #1 due
Homework #1: Read and comment on Bell (1999). Due: 1/16 (Thu), in class. Bring a printed or written copy. See the lecture note for details.
Week 3 (1/27-): MDL and MAP
  • Ballard Chapter 2.
Homework #2 (due 2/4): Read Knill et al. (1996) and Jepson and Feldman (1996) and answer these questions:
  1. What is the difference between a Bayesian observer and a Bayesian theorist?
  2. Can MDL be a Bayesian theorist?
  3. If you have a hierarchy of MDLs, what would be the potential problems?
Week 4 (2/3-): Efficiency in data representation
  • Efficiency in data representation (Ballard Chapter 4)
  • HW #2 due
  • Knill, Kersten, and Yuille (1996).
  • Jepson and Feldman (1996).
 
Week 5 (2/10-): Learning
  • Supervised learning (Ballard Chapter 8)
  • Unsupervised learning (Ballard Chapter 9)
  • Dayan (1999).
  • Barlow (1982).
Homework #3: PCA exercise -- see the lecture slide for details (due: 2/18 Tue in class).
Week 6 (2/17-): Computationalism vs. Active Approach
  • Searle (2002; Chapter 7): sorry for the late announcement. If you don't have time to read it, just come by to the class.
  • HW #3 due
  • Arbib (2003) on Schema Theory
  • Arbib (1996): will be handed out.
Homework #4: Read Searle (2002; Chapter 7) and answer: how are (1) observer-independence, (2) intrinsic intentionality, and (3) causal reality related to each other, and speculate on how activeness, or action may address these issues? (one paragraph each: due 3/6).

Paper presentation topic submission due 3/6/02 (Thu): submit (1) project team, (2) paper citation, and (3) preferred presentation date (3/17--4/4).
Week 7 (2/24-): Analogy
  • Choe (2003).
  • Hill and Tononi (2002): optional.
  • Reading: Kanerva (1998; two papers),
Optional (but highly recommended) reading: Stafford (1999), chapters 4 and 5.
Week 8 (3/3-):
  • Thalamus
  • Self-organization in complex systems
  • Hill and Tononi (2002)
  • Guillery and Sherman (2002): Optional
  • Choe (2003): xppaut simulation demo
  • Project: potential topics and introduction to existing tools.
  • Langlois and Garrouste (1997)
  • Paper presentation topic due
  • HW #4 due
 
Week 9 (3/10-): Spring Break No Classes
Week 10 (3/17-): Paper Presentation
  • Paper presentation #1 : Barani/Agustin (Rolls 1999; Rolls and Baylis 1994)
  • Paper presentation #2 : Yingwei (Blakeslee and McCourt 1999)
  • Field 1987.
  • Project/Term Paper proposal due
 
Week 11 (3/24-): Paper Presentation
  • Paper presentation #3 : Chris (Song et al. 2000: was Pearl 2001); Kuncara (Prescott et al. 2002).
  • Paper presentation #4 : Jackie (Demiris and Hayes 2001), Sejong (Bell and Sejnowski 1997; was Gentner and Markman 2003)
  Homework #5: Choe (2003) Thalamocortical model experiments (due 4/3)
Week 12 (3/31-): Paper Presentation

Imitation

  • Arbib (2003) on Imitation (in course packet)
  • Demiris and Hayes (1997)
  • Optional: Rizzolatti (2001)
  • Note: regular lecture.
  • Paper presentation #5 : Subru (Geisler et al. 2001), Kumar (Ernst and Banks 2002)
  • HW #5 due
 
Week 13 (4/7-): Multi-modal / Sensory-Motor Integration
  • Churchland et al. (1994).
  • HW #6: submit comments/questions after reading the paper above.
  • Brooks et al. (1998)
  • Cohen and Beal (2000)
 
Week 14 (4/14-): Natural Scene Statistics, Self-organization, Structure, and Function
  • Lee and Choe (2003)
  • HW #7: submit comments/questions after reading the paper above.
  • Choe and Miikkulainen (2002)
 
Week 15 (4/21-): Project Presentation
  • Regular lecture (topic TBA)
  • Term Project postponed to 5/2.
  • Presentation #1: Yingwei (Various visual illusions); Presentation #2: Agustin/Barani (Multimodal integration in the olafactory system).
  • Last class of the semester
 
Week 16 (4/28-) * we're here * Project Presentation
  • No class (dead day)
  • FRIDAY: project presentation (3-5pm)
Project presentation: 5/2 (3-5pm); Final grades due: Monday 5/12.

Reading List

Assignments

  1. See the weekly schedule.
  2. Homework #3: data and code.
  3. Homework #5: skeleton code.
  4. Term Project description is now online (2/26).
Last Updated :