An international project aimed at increasing the efficiency and portability of artificial intelligence (AI) technology has been launched with the University of Strathclyde as a partner.
The ChipAI Project is developing a nanoscale photonics-enabled technology capable of delivering compact, high-bandwidth and energy efficient neuromorphic (brain-like) central processing units (CPUs).
Nanostructures
This is to be achieved through the use of low-dimensional semiconductor nanostructures embedded in ultra-small cavities, 100 times smaller than conventional devices, for efficient light confinement, emission and detection of spiking neuron-like signals, similar to the signalling found in the brain. It will develop proof-of-principle of nanophotonic components and chips for AI applications.
The three-year academic and industrial partnership project has secured funding of nearly €3.9 million (£3.354 million) from Horizon 2020, the EU framework programme for research and innovation. It is led at INL – International Iberian Nanotechnology Laboratory, in Braga, Portugal.
Dr Antonio Hurtado, a Senior Lecturer in Strathclyde’s Institute of Photonics, is the University’s lead on the project. He said: “In the same way that the internet has revolutionised our society, the rise of Artificial Intelligence that can learn without the need of explicit instructions is transforming our lives.
“AI uses brain inspired neural network algorithms powered by computers. However, current central processing units are extremely energy-inefficient in implementing these tasks; this represents a major bottleneck for scalable and portable AI systems.
“Reducing the energy consumption needed to emulate complex brain functions is a major challenge that ChipAI is now addressing using photonics-based technologies.
“ChipAI will not only lay the foundations for the new field of neuromorphic optical computing but will also enable new non-AI functional applications in biosensing, imaging and many other fields where masses of cheap miniaturised pulsed sources and detectors are opening up disruptive innovations.”
Key components to be developed in ChipAI include non-linear nanoscale lasers, LEDs (Light-Emitting Diodes), detectors and synaptic optical links on silicon substrates, all of which will be used to make ChipAI an economically viable technology.
This radically new architecture will be tested for neuron-like information processing, towards validation for use in artificial neural networks. This will enable the development of real-time and offline portable AI and neuromorphic, or brain-like, CPUs.
Other partners in the project are the University of Glasgow, Eindhoven University of Technology, the Faculty of Sciences (FCiencias.id) at the University of Lisbon, the University of the Balearic Islands, IQE plc and IBM Research Gmbh.
ChipAI forms part of FET Open, a Horizon 2020 programme which supports the early-stages of the science and technology research and innovation around new ideas towards radically new future technologies.
The Institute of Photonics is based in Strathclyde’s Technology and Innovation Centre. This is sited in the Glasgow City Innovation District, which is transforming the way academia, business and industry collaborate to bring competitive advantage to Scotland. The model – which is recognised for improving productivity, creating jobs and attracting inward investment in several cities around the globe – brings together researchers and high-growth firms with technology and creative start-ups, to work side-by-side in vibrant, walkable innovation communities.