Reservoir computing using photon-magnon coupling
Loïc Millet, Haechan Jeon, Bosung Kim, Biswanath Bhoi, Sang‐Koog Kim
Abstract
The current demand for large-volume data processing has led to the emergence of brain-inspired devices and algorithms, such as reservoir computing (RC), a promising computational framework for temporal-data processing. As inspired by a demonstration of RC using spin-torque nano-oscillators [Marković et al., Appl. Phys. Lett. 114, 012409 (2019)], we experimentally demonstrated RC using the non-linear dynamical responses of photon-magnon coupling (PMC) modes, i.e., the magnitude, phase, and frequency of the transmission spectra in a specially designed hybrid system consisting of an inverted split-ring resonator and an yttrium iron garnet film. Through the outputs decoded from the magnitude and frequency of the PMC transmission spectra, we experimentally achieved a 100% classification rate for recognitions of non-degenerate sine and square waveforms in a wide range of DC magnetic fields (DC currents) center, thanks to the extremely high signal-to-noise ratios and the non-linearity of the dynamical variables. The experimental realization of RC based on PMC can pave an alternative pathway to the development of high-performance RC devices.