Details
Neuromorphic devices aim to emulate the structure of the human brain (neurons and synapses) to develop resource efficient computing systems. Recent developments, such as IBM’s Hermes Chip and Intel’s Loihi 2, have shown strong growth but the complexity of manufacturing an increasing number of operating elements makes it challenging to scale indefinitely. To circumvent this, we will explore the use of bespoke nanoscale devices as ‘complex’ neurons that operate like a collection of standard neurons. To train networks of these devices we aim to use digital twins; these are machine learning models trained to predict the output of the physical system but can be differentiated for training.
This project will develop the use of machine learning methods, particularly neural differential equations, for predicting the dynamics of experimental systems designed for neuromorphic computing. A focus will be on how these models can be trained to learn device variations and how they affect performance metrics. It will explore the use of meta-learning to train models so that they can be adapted to new systems with relatively few data points.
Supervisor Bio
Dr Matthew Ellis is a Lecturer in Machine Learning within the Department of Computer Science. With a background in theoretical physics, he looks at how unconventional computing systems can be used to create energy efficient hardware for AI applications. He has particular interests in unconventional machine learning algorithms, computational modelling of physical systems and how paradigms from neuroscience can lead to new ways of computing. His research intersects machine learning and physics, looking to better integrate advances in both to create new paradigms for computing.
About the Department/Research Group
This role will be based within the Department of Computer Science which has a track record of research excellence; ranking 8th nationally for the quality of our research environment and with 99 percent of our research rated world-leading or internationally excellent. It is vibrant and progressive, supported by a recent multi-million pound investment in the Centre of Machine Intelligence.
The candidate will join the Bio-Inspired Machine Learning Lab, jointly led by Dr Ellis and Prof Eleni Vasilaki, and part of the wider Machine Learning Group. They join a strong interdisciplinary collaboration crossing the Computer Science and Materials Science covering both theoretical and experimental research into using magnetic materials for novel unconventional computing applications.
Candidate Requirements
- Minimum 2.1 Bachelor’s degree or Master’s degree in a relevant discipline (e.g. Physics, Maths, Computer Science or similar), or its international equivalent.
- Experience of neural networks and machine learning, with knowledge of the fundamental methods in this area.
- Be self-motivated with a keen interest in machine learning and/or computational modelling.
- Strong programming skills, ideally in Python and C/C++.
- If English is not your first language, you must have an IELTS score of 6.5 overall, with no less than 6.0 in each component.
How to Apply
Please note that this studentship is one of three projects advertised with Dr Matt Ellis and applicants should only apply for one of these studentships. Please indicate which one you are applying for on your application.
To apply for a PhD studentship, applications must be made directly to the University of Sheffield using the Postgraduate Online Application Form. Make sure you name Dr Matt Ellis as your proposed supervisor.
Information on what documents are required and a link to the application form can be found here -https://www.sheffield.ac.uk/postgraduate/phd/apply/applying
The form has comprehensive instructions for you to follow, and pop-up help is available.
You should include a short research statement that:
- outline your reasons for applying for this studentship
- explain how you would approach the research, including details of your skills and experience in the topic area
- be no longer than 3 A4 pages, include references
Funding Notes
The PhD studentship will cover standard UK home tuition fees and provide a tax-free stipend at the standard UK Research Council rate (currently £19,237 for the 2024/25 academic year) for 3.5 years. If you are an overseas student, you are eligible to apply but you must have the means to pay the difference between the UK and overseas tuition fees by securing additional funding or self-funding. Further information on International fees can be found here: https://www.sheffield.ac.uk/new-students/tuition-fees/fees-lookup