[C-8] 超伝導エレクトロニクス

[C-8-4] In-Memory Processing XNOR Gate Reconfigured by External Dc Control Currents for Binary Neural Networks using Single Flux Quantum Circuits

^○Zeyu Han, Yuki Yamanashi, Nobuyuki Yoshikawa (Yokohama National Univ.)

この講演は本会「学術奨励賞受賞候補者」の資格対象です。

Keywords:Single flux quantum circuit、In-Memory processing、XNOR gate、Binary neural network

Single flux quantum (SFQ) circuits have the characteristics of high-speed operation and low power consumption. In this paper, we propose an in-memory XNOR gate using the SFQ circuits for Binary Neural Networks. The proposed SFQ XNOR gate is based on a memory cell reconfigured by external DC control currents. This memory cell stores the weights and was added an input signal as the activation input. We have demonstrated it can complete the XNOR operation correctly in simulation.