Neuro-Inspired Signal Processing in Ferromagnetic Nanofibers

Area of Science:

• Materials Science
• Computer Engineering
• Physics

Background:

• Modern computers face bottlenecks due to separate data storage and processing units.
• Cognitive (neuromorphic) computing aims to combine these functions, inspired by the human brain, for improved speed and energy efficiency.
• Nanofiber networks offer a potential hardware solution for neuromorphic computing, preparable via methods like electrospinning.

Purpose of the Study:

• To investigate the potential of coupled semicircle fibers in nanofiber networks for neuromorphic computing applications.
• To demonstrate how magnetic domain walls can be utilized as a mechanism for data processing in artificial synapses.

Main Methods:

• Micromagnetic simulations were performed on three coupled semicircle fibers.
• Rotating magnetic fields were used as input to excite domain walls within the fibers.
• The resulting output signals were analyzed for their potential in stochastic data processing.

Main Results:

• Excitation of domain walls by rotating magnetic fields produced distinct output signals.
• The behavior of domain walls mimicked biological synaptic activity.
• The simulated nanofiber network demonstrated suitability for artificial synapse applications.

Conclusions:

• Coupled semicircle nanofiber networks can be effectively simulated for neuromorphic computing.
• Magnetic domain wall dynamics offer a viable pathway for creating artificial synapses.
• This approach holds promise for developing faster and more energy-efficient artificial neural networks.