15-386 Neural Computation

Carnegie Mellon University

Spring

Course Description

Neural Computation is an area of interdisciplinary study that seeks to understand how the brain computes to achieve natural intelligence. It seeks to understand the computational principles and mechanisms of intelligent behaviors and mental abilities -- such as perception, language, motor control, and learning -- by building artificial systems and computational models with the same capabilities. This course explores how neurons encode and process information, adapt and learn, communicate and compute at the individual level as well as at the levels of networks and systems. It will cover basic concepts in computational and neuronal modeling, neural encoding and decoding strategies, neural networks and learning, signal processing and system analysis, statistical learning and probabilistic inference. Concrete examples will be drawn from the visual system and the motor system, including deep learning and brain computer interface. Students will learn to perform quantitative analysis and perform computational experiments using Matlab. No prior background in biology is assumed. Prerequisites: 15-100, 21-120 or permission of instructor. 21-241 preferred but not required.

Course Information

Instructors	Office (Office hours)	Email (Phone)
Tai Sing Lee (Professor)	MI 115 Friday 1:30-2:30	tai@cnbc.cmu.edu (412-268-1060)
Yimeng Zhang (TA)	Monday 6-7 pm Citadel teaching area, GHC 5th floor	yimengzh@cs.cmu.edu
Jeff Helt (1/2 TA)	Tuesday 6:30-7:30 Citadel teaching area, GHC 5th floor	jhelt@andrew.cmu.edu
Shefali Umrania (1/2 TA)	Tuesday 6:30-7:30 Citadel teaching area, GHC 5th fllor	sumrania@andrew.cmu.edu

Class location and time: GHC 4211 Monday/Wednesday 1:30 p.m - 2:50 p.m.
Website: http://www.cnbc.cmu.edu/~tai/nc17.html (course info)
Blackboard: http://www.cmu.edu/blackboard/ (Both 386/686 students should use 386 BB for access of course materials and announcements. 686 additional readings are provided in 386 BB but labeled as such.

Classroom Etiquette

Turn OFF your laptop, cell phones or any other electronic devices in the classroom.

Grading Scheme

Evaluation	% of Grade
Assignments	70
Quiz 1	10
Final Exam or Quiz 2	20
Term project (required for 686)	15

Grading scheme: A: > 88 B: > 75. C: > 65.

Assignments

6 Matlab programming and mathematical assignments on dynamical models of neurons, receptive learning, neural networks, neural decoding, deep belief nets. The solution should be in pdf file, and should be submitted before class to blackboard.
Collaboration in team of two is allowed for the first two assignments.

Term Project

A 386 student can do the term project to replace one quiz or assignment grade. It will require a 6 pages written final report and a presentation to the class. Matlab codes and additional output should also be submitted as supplementary materials in a different pdf/doc file and/or matlab zip files.

Examinations

The midterm and final exam will cover materials covered in lectures. Students are encouraged to study together for these exams.

Late Policy

You will have about 2 weeks to do each homework assignment. Homework report should be type-written. You should submit a hardcopy, documenting all the answers and results in class on the due day, as well as a softcopy of write-up and the codes to Blackboard before noon on the due day. Blackboard submission after noon is considered late.
All students are allowed to collaborate in the FIRST two assignments ONLY to facilitate matlab learning. Each should submit a copy of the solution, but should list his/her partner, if any, on the first page of the report.
Each student has a total of 5 days grace period to turn in one of five homework assignments late without penalty. These five days could be for one assignment (for 5 days) or distributed among assignments, Blackboard submission after noon on the due day is considered late.
If you need to submit a late homework after using the free late days, you will receive 5 percent deduction from maximum credit per lecture the solution set is released.
No extension possible for the term project.

Final Grade Assignment

386 students: A for total grade >= 88 %, B: < 88 AND >= 75, C: < 75, >=60. F for cheating.
686 students: A for total grade >= 88 % plus a respectable term project.

Syllabus

Date	Lecture Topic	Assignments
	LEARNING AND REPRSENTATION
W 1/18	1. Introduction
M 1/23	2. Neurons	HW 1 out;
W 1/25	3. Spikes
M 1/30	4. Matlab tutorial (guest)
W 2/1	5. Synapse and plasticity	HW 1 in
M 2/6	6. Hebbian learning
W 2/8	7. System Analysis	HW 2 out
M 2/13	8. Neural codes
W 2/15	9. Sparse coding
M 2/20	10. Competitive learning
W 2/22	11. Map learning	HW 2 in, HW 3 out.
M 2/27	12. Perceptron
W 3/1	13. Hierarchy
M 3/6	14. Midterm (guest)
W 3/8	15. Deep networks	HW 4 out.
M 3/13	Midterm grade, Spring break
W 3/15	Spring break
	ASSOCIATION and INTERACTION
M 3/20	16. Perceptual inference	HW 3 due
W 3/22	17. Decision making
M 3/27	18. Memories and imagination
W 3/29	19. Mind reading	HW 4 in. HW 5 out
M 4/3	20. Associative learning
W 4/5	21. Recurrent networks (generative models)
M 4/10	22. Ensemble codes (correlation)
W 4/12	23. Brain Networks and States	HW 5 in, HW 6 out.
M 4/17	24. Concept learning
W 4/19	25. Predictive Network
M 4/24	26. Reinforcement Learning
W 4/26	27. Motor System and BCI	HW 6 in.
M 5/1	28. Neurally-inspired intelligence
W 5/3	29. Review /project presentation	Term paper
R 5/23	Final Grade due 4 p.m.

Questions or comments: contact Tai Sing Lee
Last modified: Jan 12, 2017, Tai Sing Lee