Reservoir computing hardware implementation method and device based on coupled MEMS resonator

Abstract

The invention discloses a reservoir computing hardware implementation method and device based on a coupled MEMS resonator, and the method comprises the steps: carrying out the preprocessing of a to-be-detected time sequence signal, so as to enable the to-be-detected time sequence signal to correspond to a virtual node of the coupled MEMS resonator; designing a nonlinear vibration equation of the coupled MEMS resonator, and regulating and controlling the coupled MEMS resonator to a preset nonlinear working point according to the equation; respectively detecting two signal test ends of the MEMScoupled resonator to obtain a first output signal and a second output signal corresponding to the to-be-tested signal corresponding to each moment, and feeding back the output signal corresponding tothe to-be-tested signal at the current moment to the virtual node corresponding to the next moment through a bidirectional time delay feedback loop; and performing regression training on the preset target value and the first output signal and the second output signal corresponding to the to-be-measured signal corresponding to each moment to obtain a weight coefficient required by calculation of the storage pool. According to the method, the data mapping dimension and the memory performance are enhanced, and richer nonlinear characteristics are provided for reserve pool calculation.

Description

technical field [0001] The present disclosure relates to the field of neural network computing, and is characterized in that it relates to a method and device for realizing storage pool computing hardware based on coupled MEMS resonators. Background technique [0002] Reservoir Computing (RC for short) is a neural network algorithm model improved on the basis of Recurrent Neural Network (RNN for short). The output layer is composed. During training, its input connection weights and internal connection weights are randomly generated and remain unchanged, and only the output connection weights need to be trained. RC has been widely used in many scenarios due to its simple training method and excellent performance in temporal signal prediction, speech recognition and classification tasks. [0003] Coupled micro-electromechanical systems (MEMS for short) resonators are composed of two resonators connected by mechanical or electrostatic coupling structures manufactured by micro...

AI Technical Summary

Problems solved by technology

However, this method does not optimize the design of the resonator, which leads to high power consumption of the circuit in actual application, and does not perform optimization processing such as feature extraction on the speech signal to be tested, resulting in low accuracy. The TI-46 isolated speech digital The test accuracy of the dataset is at best 78±2%, which cannot meet more complex tasks

Images