In order to fully benefit from the potential of AI, we believe it is imperative to address the fundamental challenges associated with its trustworthiness and sustainability. “In designing projects to make AI smarter and greener, we’re taking inspiration from neuroscience – the way human brains make decisions and manage risks. This principle will underpin our approach.Dr Rajendran and Professor Simeone
08 December 2022
King's engineers awarded prestigious research fellowships to make AI more trustworthy and sustainable
Dr Bipin Rajendran and Professor Osvaldo Simeone have each won Engineering and Physical Sciences Research Council (EPSRC) Open Fellowships to research next-generation AI solutions.
Two King’s engineers have been awarded research fellowships to investigate how to make Artificial Intelligence (AI) more trustworthy and environmentally friendly. Dr Bipin Rajendran and Professor Osvaldo Simeone, from the Department of Engineering, were awarded the highly competitive Engineering and Physical Sciences Research Council Open Fellowships to undertake this research.
The two scientists are motivated by some of the most pressing technological developments set to reshape our future. These include the increasing adoption of AI in areas such as finance, medicine, and engineering, the emergence of 6G communication networks and the need to engineer new hardware platforms to run increasingly sophisticated algorithms, which on today’s computing systems are consuming huge amounts of energy.
Dr Rajendran and Professor Simeone, who will continue their longstanding collaborations through the fellowships, said:
An expert in brain-inspired computing, Dr Rajendran has been awarded a £1.5 million EPSRC fellowship for his research aimed at building new algorithms with decision-making capabilities inspired by the human brain. Furthermore, he will develop novel hardware prototypes made from 2-dimensional materials consisting of a few layers of atoms, which will improve the efficiency of the technology.
To develop this cutting-edge technology, Dr Rajendran will work with world-leading industrial partners, including Intel, Arm, IMEC, Applied Materials, and Intrinsic Semiconductors. He will also collaborate with world-renowned experts from the UK academia – Professors Bashir M. Al-Hashimi (King's), Karl Friston and Alex Shluger (UCL), Themis Prodromakis (Edinburgh) and Steve Furber and Roman Gorbachev (Manchester).
Dr Rajendran said:
“I’m excited about the opportunity to holistically pursue fundamental ideas related to AI hardware and software. The goal is to create a new class of intelligent systems that can make decisions efficiently, without relying on energy-hungry cloud computing resources, while handling uncertainty in data, similar to how our brain processes information and makes decisions. I am looking forward to collaborating with some of the world's leading experts in these topics and training a new generation of thought leaders through this fellowship.”
Professor Simeone, whose expertise lies in information theory and machine learning, has received a fellowship of £1 million. With this, he will lead a research team dedicated to designing reliable and sustainable AI systems that can be accessed by end users - individuals or machines - on mobile wireless devices, while having control over the amount of information shared with the cloud and on the energy consumed for computing and communications. This technology will pave the way for a new form of AI, moving away from the centralised, resource-intensive, data-extractive dominant framework of today towards personalised, resource-efficient, and privacy-minded solutions. Professor Simeone’s project is supported by key industry partners, including Samsung AI Centre, Intel Corporation and InterDigital.
Professor Simeone said:
“I look forward to working with the project team and with world-leading experts in academia and the industry to realise our vision of a decentralised, reliable, sustainable, and private AI platform inspired by neuroscience and information-theoretic principles.”
Professor Bashir M. Al-Hashimi, Executive Dean of the Faculty of Natural, Mathematical & Engineering Sciences, said:
“These prestigious fellowships reflect the impactful research and the relevant research challenges both Professor Simeone and Dr Rajendran are working on. I wish them continuing success in their research endeavours.”
EPSRC Open Fellowships are highly competitive funding grants generally awarded to researchers who have won significant funding grants in the past or are leaders in an area of technical development. They enable applicants to focus on projects that further discovery, innovation, instrumentation/technique development or software engineering, or a combination of these, to create a positive change in the research community.
Brain-Inspired Nanosystems for Smart and Green AI
Dr Bipin Rajendran explains his EPSRC open fellowship:
“My fellowship will tackle the goal of creating smarter and greener AI from two fundamental angles. First, it will develop novel algorithms for trustworthy decision-making motivated by the free energy principle; and, second, it will pursue the novel approach of using the undesirable imperfections of nanoscale devices based on 2D materials as a resource for efficiently implementing free energy-based AI algorithms.”
Engineering Networked Machine Learning via Meta-Free Energy Minimisation
Professor Osvaldo Simeone explains his EPSRC open fellowship
“My EPSRC Fellowship addresses the problem of designing reliable, sustainable, and privately distributed AI systems consisting of wirelessly connected, personalized, AI agents. The Fellowship’s vision is inspired by the free energy principle, which views intelligence as revolving around uncertainty quantification, resource budgeting (allostasis), and decentralisation (social intelligence). This is in stark contrast to the current dominant model of AI, which is largely driven by a pursuit of accuracy via resource-intensive, data-extractive, centralized algorithms run in the “cloud”.”