Science & Technology, UK (Commonwealth Union) – A groundbreaking advancement in brain-inspired computing, harnessing the intrinsic physical characteristics of materials to drastically minimize energy consumption, is on the brink of realization, thanks to a recent study led by researchers from UCL and Imperial College London.
The approach, known as physical reservoir computing, has faced limitations due to its lack of adaptability, as the physical properties of a material may excel in specific computing tasks but fall short in others.
In a new study published in the journal Nature Materials, an international research team utilized chiral (twisted) magnets as their computational medium. By applying an external magnetic field and adjusting temperature, they discovered that the physical properties of these materials could be tailored to suit different machine-learning tasks, overcoming the previous hurdle of limited reconfigurability.
Dr. Oscar Lee, lead author of the paper from the London Centre for Nanotechnology at UCL and UCL Department of Electronic & Electrical Engineering, indicated that this work brings us a step closer to realizing the full potential of physical reservoirs to create computers that not only demand significantly less energy but also adapt their computational properties to perform optimally across various tasks, mirroring the flexibility of our brains.
Researchers of the study point out that conventional computing is associated with substantial electricity consumption, largely due to its reliance on separate units for data storage and processing. The constant shuffling of information between these units results in energy wastage and heat production. This inefficiency is particularly pronounced in machine learning, where extensive datasets are crucial for processing. The training of a single large AI model can lead to the emission of hundreds of tonnes of carbon dioxide.
Physical reservoir computing, among various neuromorphic (or brain-inspired) approaches, seeks to eliminate the requirement for distinct memory and processing units, offering more efficient data processing methods. Apart from presenting a more sustainable alternative to traditional computing, physical reservoir computing has the potential to integrate seamlessly into existing circuitry, providing additional capabilities while maintaining energy efficiency.
In a collaborative study involving researchers from Japan and Germany, the team employed a vector network analyzer to assess the energy absorption of chiral magnets at various magnetic field strengths and temperatures that ranges from -269 °C to room temperature.
The findings revealed that different magnetic phases of chiral magnets excelled in different types of computing tasks. The skyrmion phase, characterized by magnetized particles swirling in a vortex-like pattern, demonstrated potent memory capacity suitable for forecasting tasks. On the other hand, the conical phase exhibited limited memory but optimal non-linearity for transformation tasks and classification, such as distinguishing between different types of animals.
“Our collaborators at UCL in the group of Professor Hidekazu Kurebayashi recently identified a promising set of materials for powering unconventional computing. These materials are special as they can support an especially rich and varied range of magnetic textures. Working with the lead author Dr Oscar Lee, the Imperial College London group [led by Dr Gartside, Kilian Stenning and Professor Will Branford] designed a neuromorphic computing architecture to leverage the complex material properties to match the demands of a diverse set of challenging tasks. This gave great results, and showed how reconfiguring physical phases can directly tailor neuromorphic computing performance,” explained Co-author Dr Jack Gartside, of the Imperial College London.
The collaborative effort included researchers from the University of Tokyo and Technische Universität München. Funding support for the study was provided by the Leverhulme Trust, Engineering and Physical Sciences Research Council (EPSRC), Imperial College London President’s Excellence Fund for Frontier Research, Royal Academy of Engineering, the Japan Science and Technology Agency, Katsu Research Encouragement Award, Asahi Glass Foundation, as well as the DFG (German Research Foundation).
