MlGeNN: accelerating SNN inference using GPU-enabled neural networks. Issue 2 (1st June 2022)

• Record Type: Journal Article
• Authors: Turner, James Paul; Knight, James C; Subramanian, Ajay; Nowotny, Thomas
• Is Part Of: Neuromorphic computing and engineering. Volume 2:Issue 2(2022)
• Publication Date: 2022-06-01
• Subjects: machine learning -- spiking neural networks -- GPU -- ANN to SNN conversion -- convolutional neural networks -- GeNN -- ResNet; Neural networks (Computer science) -- Periodicals; Neural computers -- Periodicals; Neuromorphics -- Periodicals; 006.3
• DOI: 10.1088/2634-4386/ac5ac5
• Languages: English
• ISSNs: 2634-4386

Abstract

Abstract: In this paper we present mlGeNN—a Python library for the conversion of artificial neural networks (ANNs) specified in Keras to spiking neural networks (SNNs). SNNs are simulated using GeNN with extensions to efficiently support convolutional connectivity and batching. We evaluate converted SNNs on CIFAR-10 and ImageNet classification tasks and compare the performance to both the original ANNs and other SNN simulators. We find 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.