Reservoir computing(RC) is a machine learning approach having small learning cost and expected as the way to relieve the burden of network and realize real-time processing. In this research, we proposed accelerometer integrated RC using oscillators made of silicon whose stiffness change and confirmed learning possibility by numerical simulation considered the vibration characteristics of the fabricated device.
- classification of walking
- anomaly detection of micro vibration
- T. Mizumoto, Y. Hirai, T. Tsuchiya,
MEMS reservoir computing using frequency-modulated accelerometer
The 68th spring meeting of Japan Society of Applied Physics (JSAP, 16-19 Mar.2021), 16p-303-2