The Neural & Machine Learning group is looking to recruit multiple Postdocs and/or PhD students funded over a period of 5 years by a ERC Starting Grant, and BBSRC, MRC and EPSRC grants. You will be working at the frontier of neuroAI, contributing to a new generation of models that will transform our understanding of learning in the brain.
All our work circles around AI-driven brain-wide credit assignment. Specific lines of work are:
1. Theories of credit assignment in cortical circuits
1.1 AI-driven cortex-wide hierarchical credit assignment (Greedy, Heng et al. NeurIPS 2022; Cosyne 2023)
1.2 Self-supervised learning in cortical networks (Nejad et al. Cosyne 2023)
1.3 Transformer-like computations in neocortical circuits
2. Neuromodulation of cortex-wide credit assignment
2.1 Theory of cognitive robustness through adaptive neuromodulatory control (Filipovica, Nejad et al Cosyne 2023)
2.2 Adaptive learning via control of local circuits (Chevy et al Cosyne 2023)
2.3 Dopaminergic-cholinergic modulation of cortical networks
2.4 Theory-informed minimisation of cognitive decline in dementia, aging and injury
3. Subcortical facilitation of learning in the cortex
3.1 Unlocking credit assignment through cortico-cerebellar networks (Boven, Pemberton et al. Nature Comms 2023; Pemberton et al. bioRxiv)
3.2 Learning to predict and plan in hippocampal-cortical networks (Mohinta, Pedamonti et al. Cosyne 2022)
Potential collaborators are:
- Randy Bruno, Helen Barron, Clare Mackay, Chris Summerfield, Armin Lak, Tim Behrens and several others [Neuro @ University of Oxford, UK]
- Randy Bruno, Masud Husain [University of Oxford, UK]
- Jack Mellor, Paul Anastasiades [University of Bristol, UK]
- Adam Kepecs [Washington University, US]
- Adil Khan [Kings College London]
- Jasper Poort [Cambridge University]
- Caglar Gulcehre [Google Deepmind, UK]
- Yoshua Bengio [MILA, Canada]
Contact the PI at rui.verylastname@dpag.ox.ac.uk if you have any questions.
****Equity, diversity, inclusion (EDI) statement****
Diversity leads to new creative solutions. We strongly encourage applications from under-represented groups, such as women, racial and ethnic or any other minorities (we are also a member of ALBA, a network for diversity in neuroscience).
