Math Prerequisite Review Resources for CSC311 Intro Machine Learning

These are the math prerequisite review resources for CSC311 Introduction to Machine Learning, used at UTM and elsewhere. These resources are intended to help review the linear algebra, calculus, and probability/statistics prerequisites used in the course.

Although these resources have low production quality, they use the same notations as in the course. They are intended to be used "just-in-time", i.e., just before each lecture or module.

If you are taking CSC311 at UTM, there will be a quiz on Quercus associated with each video. Each Quercus quiz will contain between 2-4 multiple choice questions. Below are the videos corresponding to these questions that explains the background materials for each question.

• Lecture 2: Decision Trees
  • Supervised Learning Notation
  • Joint and Conditional Probability Distributions
  • Expected Value
• Lecture 3: Linear Regression
  • Linear Algebra
  • Partial Derivative
  • Gradients
  • Contour Plots
• Lecture 4: Feature Mapping & Logistic Regression
  • Vector-Valued Functions
  • The Univariate Chain Rule
  • Vectorization note: this is a long video!
• Lecture 5-6: Multi-Class Classification; Neural Networks
  • Set Convexity
  • The Multivariate Chain Rule
  • The Multivariate Chain Rule for Gradient Computation
• Lecture 7: Bias-Variance Decomposition; Maximum Likelihood and Bayesian Estimation
  • Vectorization review, again (see above)
  • Expectations and Variance
  • Likelihood
• Lecture 8-9: Naive Bayes; Algorithmic Fairness
  • Independence
  • Conditional Independence
  • Bayes Rule
• Lecture 10: Gaussian Discriminate Analysis
  • Eigenvectors and the Spectral Theorem
  • Gaussian Distributions
  • Gaussian Maximum Likelihood
• Lecture 11: Clustering; Mixture Models; Expectation-Maximization
  • Generative Modelling
  • GDA Inference

Errata

Known issues that are not yet fixed are listed here.

• In the Eigenvectors and the Spectral Theorem videe example Bv1 should be [-2,0,0], and not [-2,2,0]