Low energy neuromorphic devices based on spin wave scattering by topological spin texture – Neurowave

Artificial neural networks underpin the current revolution in artificial intelligence (AI), reshaping
how society views, processes, and produces information. However, they have led to a large rise in the global IT power consumption due to poorly optimized hardware architectures and computationally demanding algorithm that require cloud computing. The latter limited the development of numerous AI applications, in particular edge AI, which requires low power, fast and real time local computations.
This project aims at demonstrating the large potential of a compact nanomagnetic neural network hardware to achieve fast and low energy neuromorphic computing on the edge. Our approach relies on the spin wave (SW) scattering with topological spin textures like skyrmion arrays and labyrinthine domains, where the latter can be reconfigured using magnetic fields and spin currents. Signal routing and nonlinear activation are performed by SW propagation and interference while weights are stored in a non-volatile way in the spin texture. By using the reservoir computing and physical neural network paradigms, we aim to demonstrate and benchmark pattern recognition tasks with our device.
To reach this objective, we will develop novel and unexplored scientific and technical solutions covering the field of material engineering, device fabrication and characterization, metrology, algorithmic and hardware design. This includes the development of hybrid magnetic materials for SWs propagation and scattering, SW imaging using NV center microscopy, large signal electrical SW detection, SW caustic beams, manipulation of spin textures via various external stimuli, modelling and training algorithm of physical system. The proposed energy efficient hardware with increased data protection and privacy which operates without any cloud connection would open applications from smart sensors in space and harsh environment, predictive maintenance, healthcare and automotive.