EEE 485 lecture notes
15 sections234 worked examples304 exercises104 figures
These notes are generated. They are written against the official syllabus week by week, and the section numbers are the ones the syllabus names; the book itself was not read into them. They cover the common curriculum of the course, not one particular section. Before an exam, check them against your own instructor's slides; those are what bind you.
300 entries across 15 sections
01Introduction and probability review for machine learningweek 126 worked examples · 32 exercises · 7 figures02Bayesian and frequentist machine learning: posteriors, point estimates, credible and confidence intervalsweek 220 worked examples · 30 exercises · 7 figures03Linear regression and ordinary least squaresweek 322 worked examples · 32 exercises · 7 figures04Measuring performance: training and test error, the bias-variance tradeoff and cross-validationweek 419 worked examples · 33 exercises · 7 figures05Regularized regression: ridge and lassoweek 518 worked examples · 31 exercises · 7 figures06Linear regression from a Bayesian perspective: least squares, ridge and lasso as most probable coefficients, and the full posteriorweek 619 worked examples · 31 exercises · 6 figures07Generalized linear models: logistic regression, Newton-Raphson and IRLS, more than two classes, and Poisson regressionweek 725 worked examples · 33 exercises · 7 figures08Perceptron, neural networks and backpropagation: learning from mistakes, stacking neurons, and training them by gradient descentweek 820 worked examples · 30 exercises · 7 figures09PCA, ICA and blind source separation: directions of largest variance, unmixing and the ICA likelihoodweek 924 worked examples · 30 exercises · 7 figures10Clustering: K-means, mixtures of Gaussians and the EM algorithmweek 1020 worked examples · 32 exercises · 7 figures11Feature selection: ranking by correlation and mutual information, mRMR, best subset and forward stepwiseweek 1122 worked examples · 30 exercises · 7 figures12Probabilistic graphical models: directed graphs, d-separation, naive Bayes, undirected graphs and image denoisingweek 1224 worked examples · 32 exercises · 7 figures13Restricted Boltzmann machines and deep learningweek 1320 worked examples · 30 exercises · 7 figures14Reinforcement learning: returns, Markov decision processes, value iteration and Q-learningweek 1427 worked examples · 32 exercises · 7 figures15Multi-armed bandits and online learning: regret, ε-greedy, UCB and Thompson samplingweek 1520 worked examples · 30 exercises · 7 figures
Spotted something missing or wrong? tell us · share your own notes or an old exam.
Last updated .