CMPUT 466/566: Introduction to Machine Learning

Winter 2020

Instructor: Lili Mou

All lecture notes

Additional Materials (requiring UofA sign-in)

Course logistics
Course project
Cheatsheet for Math
Written Assignments
Coding Assignments

Since I release all materials on my homepage, there is no point in being added to eclass.
Manually adding every unofficial auditing student to eclass takes too much time for me.

Course format

The instructor will derive everything on the whiteboard. No slides would be used.
No official textbook is designated, but the instructor will provide lecture notes and
point to materials for further reading.

Prerequisites

Basic math (including algebra, calculus, probability theory, and statistics)
Coding skills. The students should be able to write code in at least one programming
language, although the programming assignments would be in python.

No machine learning background is required, as this could serve as the first course in machine learning.

Grading Schema

Undergraduate students:

Written assignments: 10%
Coding assignments: 10%
Small project: 10% (Do something for something)
+ 5% bonus for a non-trivial project
Mid-term (closed book): 30%
Final (closed book): 40%

Graduate students:

Written assignments: 5%
Coding assignments: 10%
Course project: 10% for doing something for something
+ 5% for a non-trivial project
Mid-term (closed book): 30%
Final (closed book): 40%

Note:

This course focuses on paper-based derivation instead of programming.
To avoid any disputation of marking, written assignments will be graded by
a binary score for each problem. 1=a serious attempt, 0=otherwise.
For coding assignments and exams, a student can appeal for the marks only due to
scientific reasons (e.g., comments are significantly, factually wrong). A student
cannot dispute for partial mark deduction if the solution has an error. (Theoretically,
no partial marks should be given for wrong answers. If there were any partial marks,
they would reflect TA's or the instructor's courtesy.)
The final letter grade will be given by some cut-off based on numerical marks. Assuming
a student does reasonably well in all assignments and the course project, then the letter
grades roughly maps to the following criteria:

A+ = The student well understands lecture materials and can generalize to new problems.
A = The student well understands lecture materials but is unable to generalize to new problems.
A- = The student understands most part of the lecture materials, but a few details are missing.
B+ = The student understands some part of the lecture materials, but a significant portion are missing.
B = The student has some qualitative/philosophical understanding of lecture materials, but does only a little quantitative derivation.