Machine learning techniques enable us to automatically extract features from data so as to solve predictive tasks, such as speech recognition, object recognition, machine translation, question-answering, anomaly detection, medical diagnosis and prognosis, automatic algorithm configuration, personalisation, robot control, time series forecasting, and much more. Learning systems adapt so that they can efficiently solve new tasks related to previously encountered tasks.
We will cover two (mostly unreated) themes in this course.
Unsupervised learning aims to discover latent structure in an input signal where no output labels are available, an example of which is grouping news articles based on the topics they cover. Students will learn algorithms which underpin many popular machine learning techniques, as well as developing an understanding of the theoretical relationships between these algorithms. The practicals will concern the application of machine learning to a range of real-world problems.
Multi-armed bandits is the classic example of sequential decision-making problem, where an agent interacting with an uncertain environment has to maximise its reward. We will cover key notions and basic algorithms such as UCB, eps-Greedy, Thompson Sampling. We will also cover algorithms that an agent may use when confronted with an adversarial environment.
Machine learning is a mathematical discipline and it is helpful to have a good background in linear algebra, calculus, probability and algorithms. You should have taken courses on linear algebra, probability, statistics and multivariate calculus. If you need to brush up on these concepts please refer to the links provided on the resources tab.
Most material related to unsupervised learning is covered in the book by Murphy, copies of which are available in various libraries and online access is available through the Bodleian library. Apart from the excellent book by Cesa-Bianchi and Lugosi, the material on multi-armed bandits is covered in several online books listed below.
- (Mur) K. P. Murphy. Machine Learning: A Probabilistic Perspective. MIT Press 2012. (Electronic copy available through the Bodleian library.)
- (CL) N. Cesa-Bianchi and G. Lugosi. Prediction, Learning and Games. Cambridge University Press 2006.
- (Sli) Alex Slivkins. Introduction to Multi-Armed Bandits. [link]
- (BC) S. Bubeck and N. Cesa-Bianch. Regret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems. [link]