Postgraduate Course: Deep Learning (MATH11269)
Course Outline
| School | School of Mathematics |
College | College of Science and Engineering |
| Credit level (Normal year taken) | SCQF Level 11 (Postgraduate) |
Availability | Not available to visiting students |
| SCQF Credits | 10 |
ECTS Credits | 5 |
| Summary | The course will introduce students to neural networks and deep learning models, focusing on underlying methods and relation to statistical and mathematical models. Students will understand specific architectures used for different types of data and tasks and be aware of practical problems and how to mitigate them. This course will seek to give a practical introduction to these techniques, backed up by using suitable libraries in python to apply them to a variety of datasets. |
| Course description |
The precise set of topics may change slightly from year to year.
Topics may include:
- Introduction to neural networks and relation to statistical and mathematical models, including their mathematical structure and role of loss functions and activation functions.
- Principles of training by gradient descent, along with practical training problems (e.g. vanishing gradients, regularization, normalization).
- Uncertainty quantification strategies, such as ensembles and Bayesian approaches.
- Approximation theory for neural networks.
- Specific network structures for particular tasks, including convolutional networks for images and neural networks for sequences, such as recurrent networks for time series and large language models for text data.
- Generative modelling and it goals (e.g. generating data, density estimation, structure discovery, imputation), and specific examples of deep generative models such as variational autoencoders, generative adversarial networks, and/or normalizing flows.
- Interpretation of neural networks, including saliency maps, adversarial examples, influential instances, and ethical implications of AI.
|
Course Delivery Information
| Not being delivered |
Learning Outcomes
On completion of this course, the student will be able to:
- Explain the foundations of neural network models and their relationship to statistical and mathematical models.
- Describe and apply basic training algorithms for neural networks and recognise common practical challenges.
- Explain the key principles underlying deep learning.
- Describe and compare neural network architectures for structured data, such as images, time series, and text.
- Use established Python libraries to implement neural networks in practice.
|
Reading List
- Probabilistic Machine Learning: An Introduction (2022), K Murphy
- Deep Learning: Foundations and Concepts (2024), Bishop
- Deep Learning (2017), I Goodfellow et al
- Probabilistic Machine Learning: Advanced Topics (2022), K Murphy
- Interpretable ML (2025), Molnar
- Neural Networks and Deep Learning, M Nielson (2019)
- Machine Learning for Inverse Problems and Data Assimilation, Bach et al (2025) |
Additional Information
| Graduate Attributes and Skills |
The course will embed critical thinking, problem solving, and data and digital literacy into the course. Through additional exercises and workshops, student will utilize skills, such as curiosity, collaboration, communication, reflection, inclusivity, and adaptivity. The exam will assess individuality. |
| Keywords | Deep learning,neutral networks |
Contacts
| Course organiser | Dr Sara Wade
Tel: (0131 6)50 5085
Email: sara.wade@ed.ac.uk |
Course secretary | Miss Kirstie Paterson
Tel:
Email: Kirstie.Paterson@ed.ac.uk |
|
|
University Homepage