Each row is one lecture, grouped by part of the course. The slide handouts and notebooks are from ml-teaching, which also lists all slides and all notebooks. The Shorts are also collected, in course order, on the Machine learning in 1 minute page and in a YouTube playlist.
Slides: lecture handouts (PDF) · Notebooks: worked code · Apps: interactive demos · Cheat sheet: a two-page A4 summary (PDF) · Shorts: one-minute videos, each answering one question; point at a number to see the question.
Foundations
| 0 25 Jul |
Prerequisite test An open-book test on Python, linear algebra, probability and calculus, required to enrol. |
Test |
| 1 6 Aug |
Introduction and logistics What machine learning is, how the course runs, and a tour of what current models can do. |
Slides Notebooks Transcription and local LLMs, Segmentation and localization, Prompt to image Shorts 1 2 |
| 2 7 Aug |
Conventions, metrics, classification and regression Notation for data, train and test sets, the confusion matrix, accuracy, precision and recall, and errors for regression. |
Slides Notebooks Rule-based vs ML, Confusion matrix, Dummy baselines Cheat sheet Shorts 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 |
Trees, model fit and ensembles
| 3 8 Aug |
Decision trees I Entropy, information gain, and growing a tree one greedy question at a time. |
Slides Notebooks Entropy, Real output, Real input, discrete output Cheat sheet Shorts 1 2 3 4 |
| 4 13 Aug |
Decision trees II Real-valued inputs and outputs, Gini impurity, and how depth leads to overfitting. |
Slides Notebooks Overfitting and underfitting Cheat sheet Shorts 1 2 3 |
| 5 14 Aug |
Bias and variance I; cross-validation Underfitting and overfitting, the bias-variance decomposition, k-fold cross-validation and tuning hyperparameters. |
Slides Bias and variance I, Cross-validation Notebooks Tuning hyperparameters, Bias-variance analysis, Bias-variance charts Cheat sheets Bias and variance, Cross-validation and hyperparameters Shorts 1 2 3 4 5 6 |
| 6 20 Aug |
Ensemble learning Bagging, random forests and boosting, and why a vote of many models beats one. |
Slides Notebooks Boosting explained, Feature importance, Ensemble representation Cheat sheet Shorts 1 2 3 4 5 |
| 7 21 Aug |
Decision tree pruning; bias and variance II Cutting a grown tree back, and a second look at the bias-variance trade-off. |
Slides Tree pruning, Bias and variance II Notebooks Tree pruning Cheat sheets Decision trees II, Bias and variance Shorts 1 2 |
Linear models and optimisation
| 8 22 Aug |
Maths for ML Gradients of linear and quadratic forms, rank, contour plots, and the time complexity of decision trees. |
Slides Maths for ML, Contours and gradients, Tree time complexity Notebooks Contour plots, Meshgrid and contours Cheat sheet Shorts 1 |
| 9 27 Aug |
Linear regression I Least squares, the normal equation, and what one outlier does to the fit. |
Slides Notebooks Tree pruning Cheat sheet Shorts 1 2 3 |
| 10 28 Aug |
Linear regression II The geometric view, basis functions, multicollinearity and dummy variables. |
Slides Notebooks Geometric linear regression, Basis expansion, Basis expansion 2, Dummy variables Cheat sheets Linear regression I, Linear regression II Shorts 1 2 |
| 11 29 Aug |
Gradient descent From Taylor series to the update rule, the step size, and batch, mini-batch and stochastic gradient descent. |
Slides Notebooks Taylor series, Contours and gradient descent Cheat sheet Shorts 1 2 3 4 |
| 12 3 Sep |
Movie recommendation and matrix factorisation Filling in a ratings matrix with low-rank factors, fit by alternating least squares or gradient descent. |
Slides Notebooks Movie recommendation Cheat sheet Shorts 1 |
| 13 10 Sep |
Ridge regression A penalty on large weights: the closed form, shrinkage and choosing the penalty. |
Slides Notebooks Ridge regression Cheat sheet Shorts 1 2 |
| 14 11 Sep |
Lasso regression The L1 penalty and sparse weights, subgradients, coordinate descent and feature selection. |
Slides Notebooks Lasso regression Cheat sheet Shorts 1 2 |
| 15 12 Sep |
Constrained optimisation Lagrange multipliers and the KKT conditions. |
Slides Cheat sheet |
Classification
| 16 17 Sep |
Logistic regression I The sigmoid, odds and the decision boundary, and why linear regression fails for classes. |
Slides Notebooks Why use logits, Logistic regression in Torch Cheat sheet Shorts 1 2 3 |
| 17 8 Oct |
Logistic regression II Cross-entropy from maximum likelihood, its gradient, and softmax for many classes. |
Slides Notebooks Why use logits, Logistic regression in Torch Cheat sheet Shorts 1 2 |
| 18 9 Oct |
Convexity, Hessians and IRLS Tutorial: testing convexity, Newton's method, and iteratively reweighted least squares. |
Slides Notebooks Second-order logistic regression Cheat sheet Shorts 1 |
| 19 10 Oct |
Precision-recall curves Thresholds, the precision-recall trade-off, average precision and ROC curves. |
Slides Notebooks PR curves Cheat sheet Shorts 1 2 |
Neural networks
| 20 15 Oct |
Multi-layer perceptrons Perceptrons, XOR and hidden layers, activations, and the forward pass as matrices. |
Slides Notebooks Perceptron learning algorithm Cheat sheet Shorts 1 2 3 4 5 |
| 21 16 Oct |
Automatic differentiation Computation graphs, the chain rule, and forward versus reverse mode. |
Slides Notebooks Autodiff Cheat sheet Shorts 1 2 3 |
| 22 17 Oct |
Autograd from scratch Building a small reverse-mode autograd engine in Python. |
Notebooks Autograd from scratch Cheat sheet |
| 23 22 Oct |
Next-token prediction A character-level model of names: embeddings, an MLP, and sampling. |
Slides Notebooks Names Apps Next-token demo Cheat sheet Shorts 1 |
| 24 23 Oct |
Convolutional networks I Convolution as a sliding dot product, stride, padding, pooling, weight sharing and LeNet. |
Slides Notebooks 1D CNN, Edge detection, LeNet Apps Convolution, Multi-channel convolution Cheat sheet Shorts 1 2 |
| 25 24 Oct |
Grad-CAM, equivariance and invariance Seeing which pixels a CNN relies on, and how convolution treats shifted inputs. |
Notes Apps Grad-CAM, Grad-CAM enhanced |
| 26 24 Oct |
Pretrained features with DINOv3 Using a frozen vision model's features for search, classification and regression. |
Notebooks Classification, Regression, Data Apps Visual fashion search, Patch similarity Shorts 1 |
Unsupervised and reinforcement learning
| 27 29 Oct |
Unsupervised learning: clustering K-means and choosing k, and clustering by density. |
Slides Apps Simple image clustering, Image clustering, Document clustering Cheat sheet Shorts 1 2 3 |
| 28 30 Oct |
Principal component analysis Tutorial: directions of maximum variance, eigenvectors of the covariance, and explained variance. |
Slides Notebooks PCA 1, PCA 2 Cheat sheet |
| 29 31 Oct |
Reinforcement learning Agents and rewards, returns, value functions, Q-learning and exploration. |
Slides Apps Flappy Bird, RL demo, Exploration Cheat sheet Shorts 1 2 |
Neighbours and support vector machines
| 30 6 Nov |
K-nearest neighbours Distance metrics, choosing k, the curse of dimensionality and approximate search. |
Slides KNN, Approximate KNN Notebooks Curse of dimensionality Cheat sheet Shorts 1 2 3 4 5 |
| 31 12 Nov |
Support vector machines I The maximum-margin classifier, its dual, and support vectors. |
Slides Notebooks SVM, SVM with CVXOPT Cheat sheet Shorts 1 2 |
| 32 13 Nov |
Support vector machines II: kernels Feature maps, the kernel trick, and polynomial and RBF kernels. |
Slides Notebooks Understanding kernels, Kernels, Primal and dual Cheat sheet Shorts 1 |
| 33 14 Nov |
Support vector machines III: soft margin Slack variables, the C parameter and the hinge loss. |
Slides Notebooks Soft margin Cheat sheet Shorts 1 |
Also in the series
Topics with a cheat sheet or Shorts that were not separate lectures in 2025.
|
Naive Bayes and Bayesian networks Bayes' rule for classification, conditional independence, and smoothing counts. |
Cheat sheet Shorts 1 2 |
|
Time series and sequence models Forecasting baselines, autoregression, and recurrent networks. |
Cheat sheet Shorts 1 2 |
|
Object detection Bounding boxes and how much two of them overlap. |
Shorts 1 |