An introduction to the computational neuroscience of vision. This course explores the neuroscience, as well as the computational principles and mathematical foundations that are relevant to understanding intelligent computation in biological visual systems. Fundamental concepts in signals and systems, pattern analysis, Bayesian inference, representation learning, information and coding theories will be covered in the context of sensory and perceptual processing in the visual systems. Neural anatomy and physiology of the visual system underlying visual perception and object recognition will be covered in details, with discussion from both computational and psychological perspectives. No prior background in biology or psychology is required. At least one course in computer programming is required. One course in linear algebra and one course in differential equations or probability are strongly recommended. Students will master the core concepts and techniques by engaging in a number of Matlab and mathematical exercises, and a term project on related topics. Graduate students can enroll in 15-686 for graduate credit, which requires additional reading and participation in a journal club to discuss current papers on computational neuroscience of vision. The course is appropriate for undergraduate and graduate students in psychology, neuroscience, computer science, engineering, mathematics and physics with the appropriate minimal background.
Students will
| Instructors | Office Hr | Address | Phone | |
| Tai Sing Lee (Prof) | Fri 3-5 p.m. or by appt | Mellon Inst. Rm 115 | tai@cs.cmu.edu | x8-1060 |
| Yang Gu (TA) | Wed 3:30-4:30 | Wean 3702 | guyang@cs.cmu.edu | x8-8911 |
| Evaluation | Points of Grade |
|---|---|
| Homework | 50(50) |
| Term Project | 25 (25) |
| Final Exam | 20 (20) |
| Project presentation | 5 bonus (5 required) |
| Class participation | 5 (5) |
| Journal Club | (15) |
You will have 1 to 2 weeks to do each assignment. The homework assignments are intended to introduce you to some basic techniques and to prepare you for the term project research. Homework report should be type-written (and submitted as pdf file). You may collaborate with one partner on some homework (as stipulated in the assignment). One submission per group is required in collaborative project.
General guidelines: Each student is required to do an independent project. The project can be (1) an critical literature review of a particular problem, coupled with some simulation investigation; (2) a computational model to solve a particular problem; (3) a psychophysical experiment to test a model; (4) a study invovling the quantitative analysis of neural data. You are encouraged to work together with the professor and/or the TA to come up with a specific project, by the time of midterm. You are expected to write up a paper on your research. If you present orally to your classmates at the end of the semester, you will gain up to 5 points as bonus credit.
| Date | Lecture Topic | Relevant Reading | Assignments |
|---|---|---|---|
| T 1/16 | 1. What does it mean to see ? | ||
| R 1/18 | 2. Overview of the visual nervous system | ||
| T 1/23 | 3: Basic anatomy, physiology | Problem Set 1 out | |
| R 1/25 | 4. Multi-resolution Laplacian analysis | ||
| T 1/30 | 5. How does a neuron work? | ||
| R 2/1 | 6. Synapses and system response | ||
| T 2/6 | 7. Fourier analysis | Problem Set 1 due. Problem Set 2 out | |
| R 2/8 | 8. System Analysis | ||
| T 2/13 | 9. Adaptation and contrast normalization | ||
| R 2/15 | 10. V1 physiology and Gabor functions | ||
| T 2/20 | 11. Wavelet analysis and texture perception | Problem Set 2 due. Problem Set 3 out | |
| R 2/22 | 12. Hebbian learning and principal components | ||
| T 2/27 | 13. Information theory and coding | ||
| R 3/1 | 14. Independent component analysis | ||
| T 3/6 | 15. Natural Scene statistics | Problem Set 3 due. Problem Set 4 out | |
| R 3/8 | 16. Edge detection and image segmentation | Midterm grade due | |
| T 3/13 | Spring break | ||
| R 3/15 | Spring break | ||
| T 3/20 | 16. Scene statistics and Contour grouping | ||
| R 3/22 | 17. Markov random field for segmentation | ||
| T 3/27 | 18. MRF and Belief propagation | Problem Set 3 due extension | |
| R 3/29 | 19. Scene recognition | Problem Set 4 due. Problem Set 5 out | |
| T 4/3 | 20. Feature selection (boosting, SIFT) | ||
| R 4/5 | 21. Recognition hierarchy | ||
| T 4/10 | 23. Attention and context | ||
| R 4/12 | 24: MT and motion Physiology | Problem Set 5 due | |
| T 4/17 | 25. Computational modeling of motion | ||
| R 4/19 | No Class/Spring Carnival | ||
| T 4/24 | 26. Unsupervised object learning (Tom Stepleton) | ||
| R 4/26 | 27. 3D scene statistics and 3D inference (Brian Potetz) | ||
| T 5/1 | 28. Project Presention | ||
| R 5/4 | 29. Project Presentation | FCE. Term project due | |
| M 5/14 | Final Exam: 8:30-11:30 a.m. | ||
| R 5/17 | Final Grade Due |