COMP3120: Introduction to Machine Learning
A first pass at the models everything else in the field builds on, with enough of the underlying maths to know why a method fails rather than just that it did. Assignments are graded on whether your evaluation methodology holds up, not on leaderboard position.
Credit units
6 units
Instructor
Prof. Daniel Okafor
Prerequisites
MATH1013 and COMP1100, or equivalent linear algebra and programming background.
Learning objectives
By the end of this course, students will be able to:
- Fit and evaluate regression, classification and clustering models
- Diagnose why a model fails, not just report that it did
- Design an evaluation methodology that would survive a reviewer's scrutiny
- Explain the maths behind a method well enough to know its assumptions
Assessment
- Problem sets20%Four problem sets mixing derivations with small implementation tasks.
- Assignment 1 — supervised learning25%Build, tune and evaluate regression and classification models on a provided dataset.
- Assignment 2 — unsupervised learning & evaluation25%Clustering and evaluation methodology, marked on the soundness of the analysis over leaderboard position.
- Final exam30%Closed-book exam covering the theoretical content.
Assignments are submitted as a report plus code; the exam is in-person and closed-book.
Timetable
Lecture: Tue 11am–12pm, Manning Clark 1
2 tutorial option(s) — pick one on the timetable page after enrolling.
- Tutorial A: Wed 9am–10am, CSIT N201
- Tutorial B: Thu 3pm–4pm, CSIT N201
4/4 courses added.