Week 1 – Welcome / Single Layer Networks (1)

https://opencourse.inf.ed.ac.uk/MLP

Week 1

Teaching week commencing: 21 September 2026
Key topics: Single-layer networks, course overview and AI engineering.

Reading guide

• Goodfellow, Bengio and Courville: Deep Learning — Chapter 1 and Sections 4.3, 5.1 and 5.7.

Lecture materials

Scripted recording: Welcome! / Single Layer Networks (1)

This recording and its accompanying slides cover broadly similar material to the live lectures. They were produced in previous years, so references to course logistics are out of date.

Slides: Lecture 01 — Introduction to MLP and Single Layer Networks (see below).

Practical preparation

There is no lab associated with Week 1. You can begin setting up your development environment by following the instructions in the course GitHub repository.

Indicative workload

The original taught-course estimate was 5 hours:

• Recorded lecture: 1 hour.
• Lecture/discussion: 1 hour.
• Independent study: 3 hours.

Independent study includes the suggested reading and practical preparation. These estimates are guidelines; adjust the time to suit your learning needs.

Selected technical Q&A

Why does the squared-error loss include a factor of 1/2?

Differentiating a squared term produces a factor of 2. Including a factor of 1/2 in the loss cancels this factor, simplifying the gradient expression.

Multiplying the loss by a positive constant does not change its minimisers, although it does scale the gradients.

In memory of Steve Renals

A tribute to Steve Renals, founder of Machine Learning Practical.

His vision and teaching continue to shape this course. A kind, knowledgeable and much-loved mentor and teacher.