The University of Oxford is to share in a major £80 million investment to develop next-generation artificial intelligence (AI) technologies.
The funding, from the Engineering and Physical Sciences Research Council, EPSRC (part of UK Research and Innovation, UKRI), will support nine new AI research hubs in the UK, one of which will be based at the University of Oxford’s Department of Computer Science.
Led by Professor Michael Bronstein, pictured above, (DeepMind Chair in Artificial Intelligence, Department of Computer Science), the Oxford hub will focus on the mathematics and computational research which is foundational to AI, and which could unlock new, more efficient AI systems.
In particular, the hub will seek new AI breakthroughs by employing theoretical tools from currently underexplored mathematical fields.
The research programme will bring together almost 40 researchers at Oxford from across the Department of Computer Science, the Mathematical Institute and the Department of Statistics. They will focus on developing rigorous mathematical answers to four fundamental questions that underlie modern AI and machine learning systems:
- How can mathematical approaches be used to reveal hidden structures in data that can be exploited by a new class of machine learning models?
- Can mathematical tools be used to characterise machine learning models to understand when and how they work and fail?
- How can we ensure that learning and optimisation algorithms benefit from structures underlying machine learning models, to develop better, more efficient, and safer AI models?
- How can we build self-adaptive AI systems that understand their limitations, and can safely make decisions potentially affecting billions of people?
Professor Bronstein said: "In 1872, Felix Klein published his now famous Erlangen Programme – a radical new unifying theory for geometry which had a profound impact across mathematics in general, leading to the emergence of many new fields. Now is the time for an 'Erlangen Programme' for AI, based on rigorous mathematical principles that would bring better understanding of existing AI models as well as a new generation of methods."
Although based at Oxford, tcche hub will have a broad geographic coverage across the UK, bringing together leading experts in the mathematical, algorithmic, and computational fields underpinning AI and machine learning systems as well as their applications in scientific and industrial settings.
Close links with a range of industry partners will enable the research team to test promising models in applied settings, and achieve early impact.
Professor Bronstein added: "Mathematicians from a wide range of different fields – particularly geometry, topology and probability – will benefit from this gateway to new problems, applications, opportunities, and collaborations. At the same time, researchers working in machine learning and AI will benefit from the transformative influx of new expertise, new ideas, and new tools. Overall, we expect a transformative cross-fertilisation."
The hub will also contribute to developing the next generation of AI researchers, and will train at least 13 PhD students. Through obtaining a rare combination of knowledge and skills in the mathematical and algorithmic foundations of AI as well as its applications, these will be well-placed to develop cutting-edge AI systems in the UK and beyond.
Besides the University of Oxford, the programme involves researchers at the University of Aberdeen, Durham University, Imperial College London, Queen Mary University of London, and the University of Southampton. A diverse range of leading industry partners have committed to support the new CDT over the next five years, contributing a further £2 million in total in staff time, training and secondments, and co-funding for PhD students, besides £3.25 million in data resources.
The initial partners include Adarga, BBC, BenevolentAI, Graphcore, Hylomorph, Institute of Cancer Research, Morrison’s, Ofcom, Oxford Nanopore, QinetiQ, Siemens Medical, Thales UK, and Tharsus.*
Professor Charlotte Deane, Executive Chair of EPSRC, and a Professor in Oxford University’s Department of Statistics, said: "Artificial intelligence is already transforming our world. EPSRC supports world-leading research to unlock its potential and ensure it is developed and used in an ethical and responsible way. Long-term research funding has led to revolutionary advancements that have made AI a powerful tool for many applications.
"These hubs will deliver revolutionary AI innovations and tools in sectors from healthcare to energy, smart cities and environment. They will achieve this by solving key challenges and improving our understanding of AI helping to drive the increased productivity and economic growth promised by this technology."
Further information about the new AI research hubs can be found on the UKRI website.