COMP 551 Fall 2022: Applied Machine Learning (4 credits)
Course Overview
This course covers a selected set of topics in machine learning and data mining, with an emphasis on good methods and practices for deployment of real systems. The majority of sections are related to commonly used supervised learning techniques, and to a lesser degree unsupervised methods. This includes fundamentals of algorithms on linear and logistic regression, decision trees, support vector machines, clustering, neural networks, as well as key techniques for feature selection and dimensionality reduction, error estimation and empirical validation.
Class location and time:
- Location: McINTYRE Medical Building Room 522
- Time: Tuesday, Thursday 8:35-9:55
- Duration: September 1 - December 1, 2022 (12 weeks)
Prerequisite/recommended
- Required (the bare minimum): MATH 323 or ECSE 205 or ECSE 305 or equivalent
- Strongly recommended: COMP 202 or 204 or 205. Efficient working experience in Python programming will be extremely useful for this course.
- Recommended: MATH 223 Linear Algebra, MATH 423 Applied Regression
Instructor
Yue Li <yue[dot]yl[dot]li[at]mcgill[dot]ca>Teaching Assistant
- TBD <tbd[dot]tbd[at]mail[dot]mcgill[dot]ca>
- TBD <tbd[dot]tbd[at]mail[dot]mcgill[dot]ca>
- TBD <tbd[dot]tbd[at]mail[dot]mcgill[dot]ca>
Evaluation
- Quizzes (10%): 10 quizzes each worth 1%
- Midterm exam (20%): in-person written exam
- Assignments (40%): 4 group class assignments (3 students per group) as 4 mini-projects each worth 10%. All of the quesions are programming questions in Python
- Final exam (30%): in-person written exam
Recommended/Complementary Textbooks (all available online)
- Probabilistic Machine Learning: An Introduction (2022) by Kevin Murphy (Murphy22)
- Machine Learning: a probabilsitic perspective (2016) by Kevin Murphy (Murphy16)
- Pattern recognition and machine learning (2006) by Christopher Bishop (Bishop06)
- Deep Learning (2016) by Ian Goodfellow, Yoshua Bengio, and Aaron Courville (Goodfellow16).
Course coverage:
- Machine learning fundamentals (2 weeks):
- Objective functions
- Probabilistic models
- Model evaluation
- Regularization
- Gradient updates
- Non-parametric methods (1 week)
- K-nearest neighbor
- Classification and regression tree
- Support vector machine
- Linear methods (3 weeks):
- Naive Bayes
- Linear regression
- Logistic and multinomial regression
- Linear determinant analysis
- Unsupervised learning (3 weeks)
- Principal component analysis
- Matrix factorization
- K-means clustering
- Gaussian mixture model
- Deep learning methods (3 weeks)
- Multilayer perceptrons
- Convolutional neural networks
- Recurrent neural networks
- Autoencoder
- Graph representational learning