Cargando…

SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron

Application of deep convolutional spiking neural networks (SNNs) to artificial intelligence (AI) tasks has recently gained a lot of interest since SNNs are hardware-friendly and energy-efficient. Unlike the non-spiking counterparts, most of the existing SNN simulation frameworks are not practically...

Descripción completa

Detalles Bibliográficos
Autores principales: Mozafari, Milad, Ganjtabesh, Mohammad, Nowzari-Dalini, Abbas, Masquelier, Timothée
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Frontiers Media S.A. 2019
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6640212/
https://www.ncbi.nlm.nih.gov/pubmed/31354403
http://dx.doi.org/10.3389/fnins.2019.00625
_version_ 1783436603263287296
author Mozafari, Milad
Ganjtabesh, Mohammad
Nowzari-Dalini, Abbas
Masquelier, Timothée
author_facet Mozafari, Milad
Ganjtabesh, Mohammad
Nowzari-Dalini, Abbas
Masquelier, Timothée
author_sort Mozafari, Milad
collection PubMed
description Application of deep convolutional spiking neural networks (SNNs) to artificial intelligence (AI) tasks has recently gained a lot of interest since SNNs are hardware-friendly and energy-efficient. Unlike the non-spiking counterparts, most of the existing SNN simulation frameworks are not practically efficient enough for large-scale AI tasks. In this paper, we introduce SpykeTorch, an open-source high-speed simulation framework based on PyTorch. This framework simulates convolutional SNNs with at most one spike per neuron and the rank-order encoding scheme. In terms of learning rules, both spike-timing-dependent plasticity (STDP) and reward-modulated STDP (R-STDP) are implemented, but other rules could be implemented easily. Apart from the aforementioned properties, SpykeTorch is highly generic and capable of reproducing the results of various studies. Computations in the proposed framework are tensor-based and totally done by PyTorch functions, which in turn brings the ability of just-in-time optimization for running on CPUs, GPUs, or Multi-GPU platforms.
format Online
Article
Text
id pubmed-6640212
institution National Center for Biotechnology Information
language English
publishDate 2019
publisher Frontiers Media S.A.
record_format MEDLINE/PubMed
spelling pubmed-66402122019-07-26 SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron Mozafari, Milad Ganjtabesh, Mohammad Nowzari-Dalini, Abbas Masquelier, Timothée Front Neurosci Neuroscience Application of deep convolutional spiking neural networks (SNNs) to artificial intelligence (AI) tasks has recently gained a lot of interest since SNNs are hardware-friendly and energy-efficient. Unlike the non-spiking counterparts, most of the existing SNN simulation frameworks are not practically efficient enough for large-scale AI tasks. In this paper, we introduce SpykeTorch, an open-source high-speed simulation framework based on PyTorch. This framework simulates convolutional SNNs with at most one spike per neuron and the rank-order encoding scheme. In terms of learning rules, both spike-timing-dependent plasticity (STDP) and reward-modulated STDP (R-STDP) are implemented, but other rules could be implemented easily. Apart from the aforementioned properties, SpykeTorch is highly generic and capable of reproducing the results of various studies. Computations in the proposed framework are tensor-based and totally done by PyTorch functions, which in turn brings the ability of just-in-time optimization for running on CPUs, GPUs, or Multi-GPU platforms. Frontiers Media S.A. 2019-07-12 /pmc/articles/PMC6640212/ /pubmed/31354403 http://dx.doi.org/10.3389/fnins.2019.00625 Text en Copyright © 2019 Mozafari, Ganjtabesh, Nowzari-Dalini and Masquelier. http://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
spellingShingle Neuroscience
Mozafari, Milad
Ganjtabesh, Mohammad
Nowzari-Dalini, Abbas
Masquelier, Timothée
SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron
title SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron
title_full SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron
title_fullStr SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron
title_full_unstemmed SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron
title_short SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron
title_sort spyketorch: efficient simulation of convolutional spiking neural networks with at most one spike per neuron
topic Neuroscience
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6640212/
https://www.ncbi.nlm.nih.gov/pubmed/31354403
http://dx.doi.org/10.3389/fnins.2019.00625
work_keys_str_mv AT mozafarimilad spyketorchefficientsimulationofconvolutionalspikingneuralnetworkswithatmostonespikeperneuron
AT ganjtabeshmohammad spyketorchefficientsimulationofconvolutionalspikingneuralnetworkswithatmostonespikeperneuron
AT nowzaridaliniabbas spyketorchefficientsimulationofconvolutionalspikingneuralnetworkswithatmostonespikeperneuron
AT masqueliertimothee spyketorchefficientsimulationofconvolutionalspikingneuralnetworkswithatmostonespikeperneuron