mlGeNN: accelerating SNN inference using GPU-enabled neural networks

In this paper we present mlGeNN – a Python library for the conversion of arti?cial neural networks (ANNs) speci?ed in Keras to spiking neural networks (SNNs). SNNs are simulated using GeNN with extensions to e?ciently support convolutional connectivity and batching. We evaluate converted SNNs on CIFAR-10 and ImageNet classi?cation tasks and compare the performance to both the original ANNs and other SNN simulators. We ?nd that performing inference using a VGG-16 model, trained on the CIFAR-10 dataset, is 2.5× faster than BindsNet and, when using a ResNet-20 model trained on CIFAR-10 with FewSpike ANN to SNN conversion, mlGeNN is only a little over 2× slower than TensorFlow.