• 文献检索
  • 文档翻译
  • 深度研究
  • 学术资讯
  • Suppr Zotero 插件Zotero 插件
  • 邀请有礼
  • 套餐&价格
  • 历史记录
应用&插件
Suppr Zotero 插件Zotero 插件浏览器插件Mac 客户端Windows 客户端微信小程序
定价
高级版会员购买积分包购买API积分包
服务
文献检索文档翻译深度研究API 文档MCP 服务
关于我们
关于 Suppr公司介绍联系我们用户协议隐私条款
关注我们

Suppr 超能文献

核心技术专利：CN118964589B侵权必究
粤ICP备2023148730 号-1Suppr @ 2026

文献检索

告别复杂PubMed语法，用中文像聊天一样搜索，搜遍4000万医学文献。AI智能推荐，让科研检索更轻松。

立即免费搜索

文件翻译

保留排版，准确专业，支持PDF/Word/PPT等文件格式，支持 12+语言互译。

免费翻译文档

深度研究

AI帮你快速写综述，25分钟生成高质量综述，智能提取关键信息，辅助科研写作。

立即免费体验

SpykeTorch：每个神经元最多一个脉冲的卷积脉冲神经网络的高效模拟

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

作者信息

Mozafari Milad, Ganjtabesh Mohammad, Nowzari-Dalini Abbas, Masquelier Timothée

机构信息

Department of Computer Science, School of Mathematics, Statistics, and Computer Science, University of Tehran, Tehran, Iran.

CERCO UMR 5549, CNRS - Université Toulouse 3, Toulouse, France.

出版信息

Front Neurosci. 2019 Jul 12;13:625. doi: 10.3389/fnins.2019.00625. eCollection 2019.

DOI:10.3389/fnins.2019.00625
PMID:31354403
原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC6640212/
Abstract

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.

摘要

深度卷积脉冲神经网络（SNN）在人工智能（AI）任务中的应用近来备受关注，因为SNN对硬件友好且节能。与非脉冲神经网络不同，现有的大多数SNN模拟框架在处理大规模AI任务时实际效率不够高。在本文中，我们介绍了SpykeTorch，一个基于PyTorch的开源高速模拟框架。该框架使用每个神经元最多一个脉冲和秩次编码方案来模拟卷积SNN。在学习规则方面，既实现了脉冲时间依赖可塑性（STDP），也实现了奖励调制STDP（R-STDP），但其他规则也能轻松实现。除了上述特性外，SpykeTorch具有高度通用性，能够重现各种研究的结果。所提出框架中的计算基于张量，完全由PyTorch函数完成，这反过来又带来了在CPU、GPU或多GPU平台上运行时即时优化的能力。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/cd2cb05ef91f/fnins-13-00625-g0013.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/ff4b0efb1c13/fnins-13-00625-g0001.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/b2c7d50a89f0/fnins-13-00625-g0002.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/1bc4079613c5/fnins-13-00625-g0003.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/0918d501a3a5/fnins-13-00625-g0004.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/c564ec669e22/fnins-13-00625-g0005.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/59128ba67361/fnins-13-00625-g0006.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/50539f5902dd/fnins-13-00625-g0007.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/6ee5ed91aad0/fnins-13-00625-g0008.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/3f7a9122d9fb/fnins-13-00625-g0009.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/7e926c2fdd82/fnins-13-00625-g0010.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/f5783899815b/fnins-13-00625-g0011.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/b9392b2294e8/fnins-13-00625-g0012.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/cd2cb05ef91f/fnins-13-00625-g0013.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/ff4b0efb1c13/fnins-13-00625-g0001.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/b2c7d50a89f0/fnins-13-00625-g0002.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/1bc4079613c5/fnins-13-00625-g0003.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/0918d501a3a5/fnins-13-00625-g0004.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/c564ec669e22/fnins-13-00625-g0005.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/59128ba67361/fnins-13-00625-g0006.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/50539f5902dd/fnins-13-00625-g0007.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/6ee5ed91aad0/fnins-13-00625-g0008.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/3f7a9122d9fb/fnins-13-00625-g0009.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/7e926c2fdd82/fnins-13-00625-g0010.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/f5783899815b/fnins-13-00625-g0011.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/b9392b2294e8/fnins-13-00625-g0012.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/4111/6640212/cd2cb05ef91f/fnins-13-00625-g0013.jpg

相似文献

1
SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks With at Most One Spike per Neuron.SpykeTorch：每个神经元最多一个脉冲的卷积脉冲神经网络的高效模拟
Front Neurosci. 2019 Jul 12;13:625. doi: 10.3389/fnins.2019.00625. eCollection 2019.
2
SSTDP: Supervised Spike Timing Dependent Plasticity for Efficient Spiking Neural Network Training.SSTDP：用于高效脉冲神经网络训练的监督式脉冲时间依赖可塑性
Front Neurosci. 2021 Nov 4;15:756876. doi: 10.3389/fnins.2021.756876. eCollection 2021.
3
First-Spike-Based Visual Categorization Using Reward-Modulated STDP.基于首次放电的视觉分类，使用奖励调制的 STDP。
IEEE Trans Neural Netw Learn Syst. 2018 Dec;29(12):6178-6190. doi: 10.1109/TNNLS.2018.2826721. Epub 2018 May 8.
4
STDP-based spiking deep convolutional neural networks for object recognition.基于 STDP 的尖峰深度卷积神经网络的目标识别。
Neural Netw. 2018 Mar;99:56-67. doi: 10.1016/j.neunet.2017.12.005. Epub 2017 Dec 23.
5
Training Deep Spiking Convolutional Neural Networks With STDP-Based Unsupervised Pre-training Followed by Supervised Fine-Tuning.通过基于STDP的无监督预训练和监督微调来训练深度脉冲卷积神经网络
Front Neurosci. 2018 Aug 3;12:435. doi: 10.3389/fnins.2018.00435. eCollection 2018.
6
A biologically plausible supervised learning method for spiking neural networks using the symmetric STDP rule.基于对称 STDP 规则的尖峰神经网络的生物合理有监督学习方法。
Neural Netw. 2020 Jan;121:387-395. doi: 10.1016/j.neunet.2019.09.007. Epub 2019 Sep 27.
7
An unsupervised STDP-based spiking neural network inspired by biologically plausible learning rules and connections.一种基于无监督 STDP 的尖峰神经网络，灵感来自于具有生物学合理性的学习规则和连接。
Neural Netw. 2023 Aug;165:799-808. doi: 10.1016/j.neunet.2023.06.019. Epub 2023 Jun 22.
8
Paired competing neurons improving STDP supervised local learning in Spiking Neural Networks.配对竞争神经元改善脉冲神经网络中基于STDP的监督局部学习
Front Neurosci. 2024 Jul 24;18:1401690. doi: 10.3389/fnins.2024.1401690. eCollection 2024.
9
Supervised Learning in SNN via Reward-Modulated Spike-Timing-Dependent Plasticity for a Target Reaching Vehicle.通过奖励调制的基于脉冲时间的可塑性在目标到达车辆的脉冲神经网络中进行监督学习
Front Neurorobot. 2019 May 3;13:18. doi: 10.3389/fnbot.2019.00018. eCollection 2019.
10
Competitive Learning in a Spiking Neural Network: Towards an Intelligent Pattern Classifier.尖峰神经网络中的竞争学习：迈向智能模式分类器。
Sensors (Basel). 2020 Jan 16;20(2):500. doi: 10.3390/s20020500.

引用本文的文献

1
Finger Vein Recognition Based on Unsupervised Spiking Convolutional Neural Network with Adaptive Firing Threshold.基于具有自适应激发阈值的无监督脉冲卷积神经网络的手指静脉识别
Sensors (Basel). 2025 Apr 3;25(7):2279. doi: 10.3390/s25072279.
2
Direct training high-performance deep spiking neural networks: a review of theories and methods.直接训练高性能深度脉冲神经网络：理论与方法综述
Front Neurosci. 2024 Jul 31;18:1383844. doi: 10.3389/fnins.2024.1383844. eCollection 2024.
3
Advancing brain-inspired computing with hybrid neural networks.

本文引用的文献

1
Deep learning in spiking neural networks.深度学习在尖峰神经网络中的应用。
Neural Netw. 2019 Mar;111:47-63. doi: 10.1016/j.neunet.2018.12.002. Epub 2018 Dec 18.
2
BindsNET: A Machine Learning-Oriented Spiking Neural Networks Library in Python.BindsNET：一个面向机器学习的Python脉冲神经网络库。
Front Neuroinform. 2018 Dec 12;12:89. doi: 10.3389/fninf.2018.00089. eCollection 2018.
3
Deep Learning With Spiking Neurons: Opportunities and Challenges.基于脉冲神经元的深度学习：机遇与挑战。
利用混合神经网络推动受脑启发的计算。
Natl Sci Rev. 2024 Feb 26;11(5):nwae066. doi: 10.1093/nsr/nwae066. eCollection 2024 May.
4
DCENet-based low-light image enhancement improved by spiking encoding and convLSTM.基于脉冲编码和卷积长短期记忆网络改进的基于DCENet的低光照图像增强方法
Front Neurosci. 2024 Mar 5;18:1297671. doi: 10.3389/fnins.2024.1297671. eCollection 2024.
5
SHIP: a computational framework for simulating and validating novel technologies in hardware spiking neural networks.SHIP：一种用于模拟和验证硬件脉冲神经网络中的新技术的计算框架。
Front Neurosci. 2024 Jan 8;17:1270090. doi: 10.3389/fnins.2023.1270090. eCollection 2023.
6
On computational models of theory of mind and the imitative reinforcement learning in spiking neural networks.关于心理理论的计算模型和尖峰神经网络中的模仿强化学习。
Sci Rep. 2024 Jan 23;14(1):1945. doi: 10.1038/s41598-024-52299-7.
7
SpikingJelly: An open-source machine learning infrastructure platform for spike-based intelligence.SpikingJelly：一个用于基于尖峰的智能的开源机器学习基础架构平台。
Sci Adv. 2023 Oct 6;9(40):eadi1480. doi: 10.1126/sciadv.adi1480.
8
From Brain Models to Robotic Embodied Cognition: How Does Biological Plausibility Inform Neuromorphic Systems?从脑模型到具身认知机器人：生物学合理性如何为神经形态系统提供信息？
Brain Sci. 2023 Sep 13;13(9):1316. doi: 10.3390/brainsci13091316.
9
Spiking CMOS-NVM mixed-signal neuromorphic ConvNet with circuit- and training-optimized temporal subsampling.具有电路和训练优化时间下采样的尖峰CMOS-NVM混合信号神经形态卷积网络
Front Neurosci. 2023 Jul 18;17:1177592. doi: 10.3389/fnins.2023.1177592. eCollection 2023.
10
SNS-Toolbox: An Open Source Tool for Designing Synthetic Nervous Systems and Interfacing Them with Cyber-Physical Systems.SNS工具包：用于设计合成神经系统并使其与网络物理系统接口的开源工具。
Biomimetics (Basel). 2023 Jun 10;8(2):247. doi: 10.3390/biomimetics8020247.
Front Neurosci. 2018 Oct 25;12:774. doi: 10.3389/fnins.2018.00774. eCollection 2018.
4
First-Spike-Based Visual Categorization Using Reward-Modulated STDP.基于首次放电的视觉分类，使用奖励调制的 STDP。
IEEE Trans Neural Netw Learn Syst. 2018 Dec;29(12):6178-6190. doi: 10.1109/TNNLS.2018.2826721. Epub 2018 May 8.
5
Event-Based, Timescale Invariant Unsupervised Online Deep Learning With STDP.基于事件的、时间尺度不变的带STDP的无监督在线深度学习
Front Comput Neurosci. 2018 Jun 14;12:46. doi: 10.3389/fncom.2018.00046. eCollection 2018.
6
Spatio-Temporal Backpropagation for Training High-Performance Spiking Neural Networks.用于训练高性能脉冲神经网络的时空反向传播
Front Neurosci. 2018 May 23;12:331. doi: 10.3389/fnins.2018.00331. eCollection 2018.
7
Unsupervised Feature Learning With Winner-Takes-All Based STDP.基于胜者全得的脉冲时间依赖可塑性的无监督特征学习
Front Comput Neurosci. 2018 Apr 5;12:24. doi: 10.3389/fncom.2018.00024. eCollection 2018.
8
SuperSpike: Supervised Learning in Multilayer Spiking Neural Networks.超级脉冲：多层脉冲神经网络中的监督学习
Neural Comput. 2018 Jun;30(6):1514-1541. doi: 10.1162/neco_a_01086. Epub 2018 Apr 13.
9
STDP-based spiking deep convolutional neural networks for object recognition.基于 STDP 的尖峰深度卷积神经网络的目标识别。
Neural Netw. 2018 Mar;99:56-67. doi: 10.1016/j.neunet.2017.12.005. Epub 2017 Dec 23.
10
Supervised Learning Based on Temporal Coding in Spiking Neural Networks.基于脉冲神经网络中时间编码的监督学习。
IEEE Trans Neural Netw Learn Syst. 2018 Jul;29(7):3227-3235. doi: 10.1109/TNNLS.2017.2726060. Epub 2017 Aug 1.