Modules are mainly delivered via lectures and/or problem classes. They take place at University Park throughout the Autumn and Spring semesters. During the Summer semester students undertake an individual research project.
Most module content directly combines insights in brain function with relevant computational approaches. Many modules are supplemented with coursework and tutorials.
Computational Cognitive Psychology
This module teaches you cognitive psychology but also how cognition be understood in computational terms, how it can be simulation and how it compares to artificial intelligence approaches.
Topics covered include:cognitive psychology, computational approaches: connectionist networks, deep nets for vision audition and language, memory networks.
- Practical Biomedical Mathematics
This module involves the application of mathematical modelling to practical problems in biology and medicine, typical of those that mathematicians and systems biologists encounter in academia and industry.
Specific projects are tackled through workshops and student-led group activities.
Students will gain experience of applying a variety of mathematical modelling approaches to a range of biomedical problems.
- Machine Learning in Science 1
The purpose of this module is to provide an introduction to the concepts and methods of modern machine learning. It will cover:
- the basics of supervised learning and unsupervised learning as applied to a variety of problems of linear and non-linear regression
- density estimation
- data generation
The modules will be a combination of fundamental concepts and hands on application to a selection of example problems.
This modules teaches you how neural processes can be understood in computational terms and can be analysed using mathematical and computational methods.
Topics covered include:
- biophysical and reduced models of neurons
- models of networks (eg Hopfield networks, ring-attractors, rate networks)
- models of synaptic plasticity and memory
- unsupervised learning
- neural coding
- visual system
- model fitting
A selection of projects provided by our research academics will be available for you to choose from. You may develop an experimental design or prepare stimuli and run a small study. Alternatively, you may evaluate existing data and interpret the results.
You can take 20 credits from the following:
- Applied Non-Linear Dynamics
This module will cover Nonlinear oscillations, including the linear stability of limit cycles (Floquet theory), the Mathieu equation, and relaxation oscillators (using geometric singular perturbation theory).
The module will conclude with a treatment of Spatially extended systems, covering pattern formation (in both PDE and integral equation models), and weakly nonlinear analysis (amplitude equations and pattern selection).
- Data Analysis for Neuroimaging
Experience a brain imaging session at our on-campus MRI centre. You will then analyse one of the data sets in further lab classes.
- Advanced Methods in Psychology
The module provides an insight into some more advanced or specialised techniques of data collection, organisation and analysis in psychological research (eg eye-tracking, EEG, fMRI, TMS, computational modeling, diary methodologies and workshops).
Lectures will include implementation of analytical procedures in for example specialised data management and statistical packages and on specialised data gathering equipment and software.
- Analytical Research Methods
A selection of workshops on advanced statistics for the neurosciences.
- The Physics of Deep Learning
This module explores the connections between models of neural networks used for machine learning applications and ideas from physics, in particular from thermal physics and statistical mechanics.
It will discuss the connections between the physics of disordered spin models and NN models of associative memory such as:
- Hopfield models
- NN models for unsupervised learning such as Boltzmann machines
- models of supervised learning such as feed-forward networks
- the relation between physical coarse-graining and convolutional NNs
It will study the problem of optimisation in rugged energy landscapes and its connection with parameter learning in deep NNs.
You will also cover connections to the physics of stochastic processes, and cover in detail numerical optimisation methods, connecting ideas from stochastic gradient descent to physical methods such as thermal annealing and parallel tempering, and sampling methods such as Monte Carlo.
- Machine Learning in Science 2
The purpose of this module is to cover more advanced topics in machine learning and artificial neural networks following Machine Learning in Science Part 1.
Topics to be covered will include:
- deep neural networks and deep supervised learning
- convolutional NNs, RNNs, GANs
- unsupervised learning, restricted Boltzmann machines, deep RBMs and autoencoders
- reinforcement learning and Markov decision processes
- cleaning data and handling large data sets
Concepts will be applied to associated small projects.
You'll examine current techniques for the extraction of useful information about a physical situation from individual and sets of images.
You'll cover a range of methods and applications, with particular emphasis being placed on the detection and identification of objects, image segmentation, pose estimation, recovery of three-dimensional shape and analysis of motion.
These problems will be approached with both traditional and modern computer vision approaches including deep learning.
The above is a sample of the typical modules that we offer but is not intended to be construed and/or relied upon as a definitive list of the modules that will be available in any given year. Due to the passage of time between commencement of the course and subsequent years of the course, modules may change due to developments in the curriculum and information is provided for indicative purposes only.