Energy-Efficient High-Accuracy Spiking Neural Network Inference Using Time-Domain Neurons
This work addresses energy efficiency and accuracy issues in neuromorphic computing for low-power AI applications, representing an incremental improvement over existing neuron designs.
The paper tackled the problem of high power consumption and nonlinearity in conventional analog voltage-domain integrate-and-fire neuron circuits for spiking neural networks, resulting in a proposed time-domain neuron circuit that achieved over 4.3x lower error rate on MNIST inference and 0.230uW power per neuron.
Due to the limitations of realizing artificial neural networks on prevalent von Neumann architectures, recent studies have presented neuromorphic systems based on spiking neural networks (SNNs) to reduce power and computational cost. However, conventional analog voltage-domain integrate-and-fire (I&F) neuron circuits, based on either current mirrors or op-amps, pose serious issues such as nonlinearity or high power consumption, thereby degrading either inference accuracy or energy efficiency of the SNN. To achieve excellent energy efficiency and high accuracy simultaneously, this paper presents a low-power highly linear time-domain I&F neuron circuit. Designed and simulated in a 28nm CMOS process, the proposed neuron leads to more than 4.3x lower error rate on the MNIST inference over the conventional current-mirror-based neurons. In addition, the power consumed by the proposed neuron circuit is simulated to be 0.230uW per neuron, which is orders of magnitude lower than the existing voltage-domain neurons.