Postdoctoral Researcher in Neuromorphic Engineering

University of Oxford; Oxford; On-site; GBP 38,000 - 48,000; Full time

Job summary

The University of Oxford invites applications for a research post to develop an implementation of predictive coding in superconducting hardware. The successful candidate will work in the Brain Network Dynamics Unit on a joint project with Great Sky, leveraging mathematical analysis and simulations to adapt the brain-inspired model for neuromorphic hardware.

Interdisciplinary collaboration with computational neuroscientists and engineers will be central, with supervision from Rafal Bogacz and

Responsibilities

• Develop a hardware implementation of predictive coding in superconducting circuits.
• Perform mathematical analysis and computer simulations to adapt the model for neuromorphic hardware.
• Collaborate with Oxford engineers and the industrial partner Great Sky under joint supervision.

Job description

Nuffield Department of Clinical Neurosciences (NDCN), Brain Network Dynamics Unit, Mansfield Road, Oxford, OX1 3TH
We are looking for an outstanding researcher for a joint project between the University of Oxford and an industrial partner, Great Sky. The postholder will develop an implementation of a model of information processing in the brain, called predictive coding, in hardware employing superconducting computational circuits. The research will employ mathematical analysis and computer simulations to investigate how the predictive coding model can be adjusted so it can be implemented in the neuromorphic hardware, while preserving the computational advantages of the model.
Predictive coding is a brain-inspired model which learns with fewer examples than traditional artificial neural networks, used in artificial intelligence (AI). Therefore, it has a potential to speed up training of AI models. However, simulating this model on current digital computers is slow, as they are not optimized for the computations which predictive coding requires. By contrast, the new hardware developed by Great Sky performs computations in a fundamentally different way, which is compatible with the requirements of the predictive coding model. Implementing predictive coding on this hardware will unlock the advantages of predictive coding, resulting in faster and more energy-efficient training of AI.
The postholder will be co-supervised by Rafal Bogacz from University of Oxford and Jeff Shainline from Great Sky. The post holder will work as a part of an interdisciplinary team including computational neuroscientists studying learning in brain networks in Oxford, and will collaborate with engineers developing neuromorphic hardware at Great Sky. The post holder will also be a member of a research group of Professor Rafal Bogacz, and based in the Brain Network Dynamics Unit, which includes an interdisciplinary community of neurobiologists, engineers and computational scientists.
For any queries regarding visa sponsorship, HR and the University's Staff Immigration Team will provide advice and assistance when the chosen candidate for the role has been identified.
This post is full time, fixed term for 2 years.