Electroencephalogram (EEG) measures electrical activity in the brain, and hence provides a medium to better understand, diagnose and treat neurological disorders and mental health conditions.
Machine learning techniques have been applied to a range of EEG-based applications as a means of decoding raw EEG into useful information, and have shown highly promising results. These applications are safety critical – misdiagnosis of a neurological disorder could have disastrous consequences for a patient. It is therefore necessary that these machine learning models have the ability to express uncertainty in the predictions they make. If uncertainty is high, further analysis can be carried out before a decision is made. This concept of uncertainty quantification is a fundamental component of safe and responsible decision-making.
To that end, this project will use a Bayesian framework for machine learning, using techniques such as Gaussian Processes and Bayesian neural networks. These techniques will be applied to EEG to improve our understanding and treatment of neurological disorders.
Brain-computer interface.
Machine learning.
Biosignal processing.
Human-machine interaction.
MEng in Bioengineering at The University of Sheffield.
Research assistant on a project using deep learning to improve an autopilot system.
Data scientist working on predictive maintenance with industrial IoT for manufacturing.
Dr Xi Chen
Dr George Stothart
