Module manager: Dr Sebastian Ordyniak
Email: s.ordyniak@leeds.ac.uk
Taught: Semester 1 (Sep to Jan) View Timetable
Year running 2026/27
COMP3910
This module is not approved as a discovery module
Optimisation problems are everywhere. They are apparent in engineering, data science, logistics, business analytics, economics, manufacturing, services. They are less visible, but play an important role as components of various systems, such as neural networks, robotics, image processing, resource usage in distributed computing, etc. This module introduces the powerful toolkit of continuous and discrete optimisation, covering theoretical aspects and applications, with the focus on applications in Artificial Intelligence.
The goals of this module are threefold: (1) to introduce fundamental methods for formalising problems as mathematical models, (2) to provide a solid foundation in understanding core optimisation methods, and (3) to give practice in applying optimisation techniques within to problems which arise as part of AI methods.
On successful completion of this module a student will have demonstrated the ability to:
1. convert problems arising in AI methods into mathematical models, select and apply appropriate computational and analytical techniques (C3, M3);
2. select appropriate solution methods, ranging in computation time and solution accuracy, suitable for actual application scenarios (C4, M4);
3. select and evaluate technical literature and other sources of information to address complex problems (C4, M4);
4. select and use practical laboratory and workshop skills to investigate complex problems and be able to comment on their limitations (C12, M12, C13, M13);
5. communicate effectively on complex engineering matters with technical and non-technical audiences, evaluating the effectiveness of the methods used (C17, M17);
6. reflect on their level of mastery of subject knowledge and skills and plan for personal development. (C18, M18).
Optimisation models:
continuous /discrete models
linear / non-linear models
LPs / ILPs / MILPs
Modelling logical implications via MILP (with links to logical inference)
Modelling via network flows (with examples arising in image processing)
Graph models (with links to neural networks)
Optimisation methods:
LP/ILP/MILP-based approaches
Alternative approaches
Exact algorithms for tractable problems (with links to image processing)
Heuristics and approximation algorithms
Branch-and-bound
Local search: Iterative Improvement, Tabu search, Simulated Annealing
Nonlinear optimisation: a brief revision of methods for solving common optimisation problems that arise in machine learning and data science, for example; least squares linear regression, gradient descent, the Newton method and enhancements (stochastic gradient descent, batched gradient descent, auto differentiation, backpropagation).
Convex optimisation with constraints: KKT conditions, the method of Lagrange Multipliers (with links to support vector machines and deep learning)
Multicriteria optimisation:
Aggregation methods,
Budgeted optimisation
Pareto dominance
| Delivery type | Number | Length hours | Student hours |
|---|---|---|---|
| Practicals | 11 | 2 | 22 |
| Lecture | 22 | 2 | 44 |
| Private study hours | 134 | ||
| Total Contact hours | 66 | ||
| Total hours (100hr per 10 credits) | 200 | ||
Check the module area in Minerva for your reading list
Last updated: 30/04/2026
Errors, omissions, failed links etc should be notified to the Catalogue Team