Jove
Visualize
联系我们
JoVE
x logofacebook logolinkedin logoyoutube logo
关于 JoVE
概览领导团队博客JoVE 帮助中心
作者
出版流程编辑委员会范围与政策同行评审常见问题投稿
图书馆员
用户评价订阅访问资源图书馆顾问委员会常见问题
研究
JoVE JournalMethods CollectionsJoVE Encyclopedia of Experiments存档
教育
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab Manual教师资源中心教师网站
使用条款与条件
隐私政策
政策

相关概念视频

Graded Potential01:19

Graded Potential

3.8K
Graded potentials are localized fluctuations in the cell membrane's electrical charge, commonly found in the dendrites of neurons. The magnitude of these potential changes depends on the strength of the initiating stimulus. In a membrane at its resting potential, a graded potential signifies a voltage shift either above -70 mV or below -70 mV.
Graded potentials fall into two categories: depolarizing and hyperpolarizing. Depolarizing graded potentials typically occur when sodium (Na+) or...
3.8K

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序
Same author

Hedgehog signaling enhances the Schwann-like and nerve repair-supportive properties of ectomesenchymal stem cells.

Frontiers in neuroscience·2026
Same author

Incorporating brain-inspired mechanisms for multimodal learning in artificial intelligence.

Science advances·2026
Same author

An online packing strategy for high volume injection in sensitivity-enhanced UPLC-MS/MS analysis of 30 veterinary drugs in beef.

Journal of chromatography. A·2026
Same author

Biologically inspired spiking diffusion model with adaptive lateral selection mechanism.

Neural networks : the official journal of the International Neural Network Society·2026
Same author

Interpretable LightGBM model for predicting postoperative gastrointestinal hemorrhage in elderly hip fracture patients: leveraging systemic inflammation and medication exposures for personalized risk stratification.

BMC geriatrics·2026
Same author

Continual learning of multiple cognitive functions with a brain-inspired temporal development mechanism.

National science review·2026

相关实验视频

Updated: Jun 21, 2025

A Simple Stimulatory Device for Evoking Point-like Tactile Stimuli: A Searchlight for LFP to Spike Transitions
07:34

A Simple Stimulatory Device for Evoking Point-like Tactile Stimuli: A Searchlight for LFP to Spike Transitions

Published on: March 25, 2014

9.9K

直接训练时间尖端神经网络,使用稀疏的代用梯度.

Yang Li1, Feifei Zhao1, Dongcheng Zhao2

  • 1Brain-inspired Cognitive Intelligence Lab, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China; School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China.

Neural networks : the official journal of the International Neural Network Society
|July 16, 2024
PubMed
概括
此摘要是机器生成的。

这项研究引入了蒙面替代梯度 (MSG) 和时间加权输出 (TWO),以改善尖端神经网络 (SNN) 训练. 这些方法通过平衡训练有效性和梯度稀疏性来提高SNN的性能.

关键词:
直接培训 直接培训 直接培训稀少的替代品梯度渐变尖神经网络的神经网络时间加权输出.

更多相关视频

Using Neuron Spiking Activity to Trigger Closed-Loop Stimuli in Neurophysiological Experiments
05:19

Using Neuron Spiking Activity to Trigger Closed-Loop Stimuli in Neurophysiological Experiments

Published on: November 12, 2019

7.0K
Optical Recording of Suprathreshold Neural Activity with Single-cell and Single-spike Resolution
08:48

Optical Recording of Suprathreshold Neural Activity with Single-cell and Single-spike Resolution

Published on: September 5, 2012

11.9K

相关实验视频

Last Updated: Jun 21, 2025

A Simple Stimulatory Device for Evoking Point-like Tactile Stimuli: A Searchlight for LFP to Spike Transitions
07:34

A Simple Stimulatory Device for Evoking Point-like Tactile Stimuli: A Searchlight for LFP to Spike Transitions

Published on: March 25, 2014

9.9K
Using Neuron Spiking Activity to Trigger Closed-Loop Stimuli in Neurophysiological Experiments
05:19

Using Neuron Spiking Activity to Trigger Closed-Loop Stimuli in Neurophysiological Experiments

Published on: November 12, 2019

7.0K
Optical Recording of Suprathreshold Neural Activity with Single-cell and Single-spike Resolution
08:48

Optical Recording of Suprathreshold Neural Activity with Single-cell and Single-spike Resolution

Published on: September 5, 2012

11.9K

科学领域:

  • 人工智能的人工智能
  • 计算神经科学是一种神经科学.
  • 机器学习 机器学习

背景情况:

  • 尖端神经网络 (SNN) 提供节能,基于事件的计算.
  • 对SNN的直接培训是具有挑战性的,因为它们的全部或没有尖端性质.
  • 替代梯度 (SG) 允许SNN训练,但可以减少稀疏性和性能.

研究的目的:

  • 为了解决由替代梯度训练减少的稀疏性引起的SNN中的性能损失.
  • 提出新的方法,以有效和稀疏的SNNs培训.
  • 提高尖端神经网络的泛化能力.

主要方法:

  • 分析直接培训问题与替代梯度的分析.
  • 引入掩盖替代梯度 (MSG) 以平衡训练有效性和梯度稀疏性.
  • 实施一个时间加权输出 (TWO) 方法用于网络解码.

主要成果:

  • 蒙面替代梯度 (MSG) 有效平衡训练和梯度稀疏性.
  • 时间加权输出 (TWO) 方法加强了正确时间步骤的重要性.
  • 结合MSG和TWO方法在实验中超越了最先进的训练技术.

结论:

  • 拟议的MSG和TWO方法为训练尖端神经网络提供了一种优越的方法.
  • 这些技术可以提高SNN的通用化和性能,同时保持理想的稀疏性.
  • 这项工作推动了SNN在神经形态硬件中的实际应用.