Course Outline
Computational Perception
COMP 546 (4 credits)
Winter 2017
TR 11:3512:55 BURN 1B23
Instructor: Professor
Michael Langer
Office:
McConnell Engineering, rm. 329
Tel:
5143983740
Email:
langer@cim.mcgill.ca
Office Hours:
TJ 12 or by appointment
Teaching Assistant (T.A.) David Bourque
Email:
david.bourque [at] mail.mcgill.ca
Office and Hours:
by appointment
Official Course Description from McGill Calendar
Computational models of visual perception and audition.
Vision problems include stereopsis, motion, focus, perspective, color.
Audition problems include source localization and recognition.
Emphasis on physics of image formation, sensory signal processing,
neural pathways and computation, psychophysical methods.
Overview
This course examines fundamental computational problems in visual and
auditory perception. Unlike traditional
perception courses offered in Psychology or Physiology departments
which emphasize neural mechanisms, this course emphasizes computational
aspects of perception.
The course consists of two main topics, namely vision and audition.
For both of these perceptual modalities, we begin by examining
the signals from the environment, namely visual and auditory images,
and the information that is contained in these images.
For vision, we consider color, shading, binocular disparity, motion, and focus.
For audition, we consider information carried by impact vs,
nonimpact sounds, echos, as well as binaural timing and intensity differences.
For both vision and audition, we then examine how images are
processed by the sensory system, using concepts and tools from linear
system theory.
For vision, we discuss retinal and cortical processing.
For audition, we discuss how sound waves are
decomposed into frequency bands by the ear and encoded by the auditory
cortex. We then examine how
properties of the environment can be inferred from the information that
is extracted from images. For vision, we consider how depth
is
estimated and for audition, we consider how depth
and direction are estimated.
Detailed Course Content and Materials
The course will consist of 24 lectures, each 80 minutes. Lecture slides, notes, and exercises
will be given as PDFs on the
public course web page . Students will be examined on this posted material. There is no
textbook for the course.
Prerequisites
There are no official prerequisites for the course. It is assumed students
can program in a high level language, at least that level of COMP 250,
and are comfortable with basic mathematics needed for an undergrad degree in computer science, in particular:
 multivariable calculus (MATH 222 or equivalent)
 linear algebra (MATH 223 or equivalent)  e.g. orthonormal basis for vector space, complex numbers
 probability (normal/Gaussian distributions and
definitions such as mean and variance, joint and conditional
probabilities).
 waves and optics (CEGEP level or PHYS 101/102).
The course will cover basic psychology and physiology of vision
and audition. It will also cover the basic tools of linear
system theory (convolution, Fourier transforms). No prior knowledge of these topics is assumed.
Evaluation
The course grade will be determined as follows:
 Four Assignments (4 x 10% = 40%)
Assignments will involve some MATLAB programming, and
questions related to the lectures or to specific research articles
related to the lectures. Students are not required to know
MATLAB prior to the course.
Policy
on lateness:
Without a medical excuse, students will be penalized by 2
course percentage points per day for handing in a late assignment, up to a maximum of 3
days late after which assignments will not be accepted.
 Exams
All exams are closed book. Students are permitted one 8x11 twosided crib sheet.
 Two midterm exams (2 x 15% = 30%)
Midterm 1 will cover roughly the first third of the course. Midterm 2 will cover roughly the second third.
Students who miss a midterm exam and have a valid (e.g. medical) reason will be eligible to write a makeup exam.
Makeup exams will be held during the final exam period.
 Final Exam (30%)
It will take place during the final exam period. It will cover
the whole course, with a much heavier weighting on the last third.
According to School of Graduate Studies
rules,
M.Sc. and Ph.D. students must achieve a grade of 65 or more to pass the
course.
Other Policies/Rules

Regrading: Mistakes can occur when grading. Not surprisingly, requests for regrading almost always
involve those mistakes in which the student receives fewer points than they deserved, rather than more. With
that in mind, note that if you wish me to regrade a question on an exam or assignment, I will do so.
However, to avoid grade ratcheting, I reserve the right to regrade other questions as well.
 Bonus points: If you contribute to the course by informing me of errors in the lecture slides, exercises, or
assignments, then I will be very appreciative and I will make a note of it. If your final grade is just below
some threshold, then your contributions may elevate your grade over that threshold.
 Final grade: There are many factors that determine your grades including how hard you work, how
talented you are in this area, how much time you have available because of other commitments, what your
academic background is, what your health situation or family situation is, etc. However, when I assign
your final course grade, I will not take these other factors into account. I assign the final grade
only based on your assignment and exam scores.
 Additional Work: Students with grades of D, F or J will not be given the opportunity to complete
additional work to upgrade their grade.
 Supplemental Exam: It will cover the material uniformly from the entire course and will replace the midterm grades.
 Cheating/Collaboration: I encourage you to discuss the assignments with each other. But no sharing
code! Your discussion should be public in the sense that anyone including me should be allowed to
listen in.
MyCourses Discussion Board
I will moderate the discussion board. Please adhere to the following:
 Be clear: Take an extra minute and make sure that what you have written makes sense.
 Be organized: Choose a suitable subject line, and stick to one topic per posting/thread in order to
allow easier indexing.
 Use the search feature to see if your question has been asked before.
 Do not email me with a technical question about the course material. Instead, post the question on
the discussion board so that everyone can benefit from the correspondence.
 If you would like your posting to be removed – for example, because you realized that your question
or comment made no sense – add a request within the thread and I will remove it. No problem.
Academic
Integrity
McGill
University values
academic integrity . Therefore, all students
must understand the meaning and consequences of cheating,
plagiarism and other academic offences under the Code of Student
Conduct and Disciplinary Procedures.