[10p-B21-1~8] Advancing the frontiers of AI electronics x Research on Systems

1:30 PM - 2:00 PM JST(4:30 AM - 5:00 AM UTC)

[10p-B21-1] Application of Analog CMOS Spiking Neural Networks to Reservoir Computing

〇Shigeo Sato¹, Satoshi Moriya¹, Hideaki Yamamoto¹ (1.RIEC, Tohoku Univ.)

Keywords:

Reservoir Computing,Spiking Neural Network,Analog CMOS Circuit

For AI processing in edge environments, low-power hardware for time-series information processing is increasingly required. This talk presents a reservoir computing system that uses an analog CMOS spiking neural network as the reservoir layer. The system combines a developed LSI chip with an FPGA and a PC. Application examples for speech and image recognition tasks will be shown, and the recognition performance, power efficiency, and potential for edge computing applications will be discussed.