An On-Chip Trainable Neuron Circuit for SFQ-Based Spiking Neural Networks

We present an on-chip trainable neuron circuit. Our proposed circuit aims at bio-inspired spike-based time-dependent data computation for training spiking neural networks (SNN). The thresholds of neurons can be increased or decreased depending on the desired application-specific spike generation rate. This mechanism is scalable and provides us with a flexible circuit structure design. We simulated the trainable neuron structure under different operating scenarios with thermal noise included. The circuits are designed and optimized for the MIT LL SFQ5ee fabrication process. For a 16-input neuron with four different threshold values, all of the circuit parameter margins are above 20% (
$\pm$
10%) with a 3G sample per second throughput.