Performance Enhancement of Occupancy Estimation through Spike-Based Reservoir
Haruto Ota, Ryuji Nagazawa, Jiaying Lin, Kien Nguyen and Hiroo Sekiya
NCN2023 IEEE Workshop on Nonlinear Circuit Networks, Dec., 2023. [pdf document]

<Abstract>

Wireless Brain-Inspired Computing (WiBIC) aims to achieve a fully wireless intelligent environment system by integrating the Internet of Things (IoT) network and Spiking Neural Network (SNN) while applying Wireless Power Transfer (WPT) technology. In the first stage of WiBIC’s application, successful learning has been demonstrated in the XOR function and occupancy estimation, establishing the feasibility of WiBIC. In this paper, we present improvements of approximately 10% in estimation accuracy for the learning model of an occupancy estimation system. This enhancement was achieved by revisiting the spike encoding method and the functionality of the reservoir to meet higher learning capabilities.