Abstract:Spiking neural networks (SNNs), as brain-inspired neural network models based on spikes, have the advantage of processing information with low complexity and efficient energy consumption. Currently, there is a growing trend to design hardware accelerators for dedicated SNNs to overcome the limitation of running under the traditional von Neumann architecture. Probabilistic sampling is an effective modeling approach for implementing SNNs to simulate the brain to achieve Bayesian inference. However, sampling consumes considerable time. It is highly demanding for specific hardware implementation of SNN sampling models to accelerate inference operations. Hereby, we design a hardware accelerator based on FPGA to speed up the execution of SNN algorithms by parallelization. We use streaming pipelining and array partitioning operations to achieve model operation acceleration with the least possible resource consumption, and combine the Python productivity for Zynq (PYNQ) framework to implement the model migration to the FPGA while increasing the speed of model operations. We verify the functionality and performance of the hardware architecture on the Xilinx Zynq ZCU104. The experimental results show that the hardware accelerator of the SNN sampling model proposed can significantly improve the computing speed while ensuring the accuracy of inference. In addition, Bayesian inference for spiking neural networks through the PYNQ framework can fully optimize the high performance and low power consumption of FPGAs in embedded applications. Taken together, our proposed FPGA implementation of Bayesian inference with SNNs has great potential for a wide range of applications, it can be ideal for implementing complex probabilistic model inference in embedded systems.

Algorithms for Fast Spiking Neural Network Simulation on FPGAs

Fast Algorithms for Spiking Neural Network Simulation with FPGAs

Spike Trains Encoding Optimization for Spiking Neural Networks Implementation in FPGA

Hardware implementation of spiking neural networks on FPGA

Enabling Efficient On-Edge Spiking Neural Network Acceleration with Highly Flexible FPGA Architectures

Real-Time Cortical Simulation on Neuromorphic Hardware

SpikeNC: an Accurate and Scalable Simulator for Spiking Neural Network on Multi-Core Neuromorphic Hardware

Efficient Implementation of Spiking Neural Networks for Inference Using Ex-Situ Training

Hardware-Software Co-optimised Fast and Accurate Deep Reconfigurable Spiking Inference Accelerator Architecture Design Methodology

A Novel Nonlinear Function Evaluation Approach for Efficient FPGA Mapping of Neuron and Synaptic Plasticity Models

Highly efficient neuromorphic learning system of spiking neural network with multi-compartment leaky integrate-and-fire neurons

An FPGA Implementation of Deep Spiking Neural Networks for Low-Power and Fast Classification

SuperNeuro: A Fast and Scalable Simulator for Neuromorphic Computing

FireFly: A High-Throughput Hardware Accelerator for Spiking Neural Networks with Efficient DSP and Memory Optimization

A Low-Cost and High-Speed Hardware Implementation of Spiking Neural Network

High Performance Simulation of Spiking Neural Network on GPGPUs

Advancements in Algorithms and Neuromorphic Hardware for Spiking Neural Networks

Spiking Neural Network Implementation on FPGA for Multiclass Classification

An FPGA implementation of Bayesian inference with spiking neural networks

GaBAN: A Generic and Flexibly Programmable Vector Neuro-processor on FPGA

Compiling Spiking Neural Networks to Neuromorphic Hardware