Overview

Machine learning is the study of algorithms that allow machines to learn patterns and make predictions based on data. Formally, it explores how a system can generalise from observed samples to unseen instances, minimising error whilst balancing constraints such as computational complexity and data availability. In essence, learning is a mathematical process that seeks to identify a hypothesis function from a class of possible functions that performs well on unseen data. In this course, we will delve deeply into the mathematical underpinnings of machine learning. Our goal is not just to implement algorithms but to rigorously understand why they work and when they are expected to fail. You will engage with mathematical concepts such as probability theory, optimisation, and linear algebra, as they are essential tools for analysing learning algorithms.

Announcement

To be announced.

Schedule

To be announced.

Lecture Material

This course is supported by several core texts that provide the essential theoretical and practical knowledge required for this course on machine learning theory. In addition to the main readings, extra materials will be recommended throughout the course for those who are curious to explore advanced topics and gain deeper insights. Below is a brief description of the primary materials we will use, along with supplementary texts available for further study. Current version of the suggested reading materials can be downloaded here.