在尖端神经网络中,基于事件的有效延迟学习
Balázs Mészáros1,2, James C Knight3, Thomas Nowotny4
1Sussex AI, School of Engineering and Informatics, University of Sussex, Brighton, United Kingdom. b.mszros@sussex.ac.uk.
Nature communications
|November 24, 2025
概括
本研究介绍了一种有效的基于事件的训练方法,用于带有延迟的尖端神经网络,提高它们对复杂任务的记忆力和准确性. 新方法比现有方法更快,使用的内存更少.
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.
背景情况:
- 尖端神经网络 (SNN) 通过稀疏的通信提供节能计算,与传统的人工神经网络 (ANN) 相反.
- 由于有状态的神经元,SNN在本质上是反复的,因此它们适合时空处理,但它们的内在记忆受到时间常数的限制.
- 延迟提供了一个强大的机制来扩展SNN中的内存.
研究的目的:
- 为SNN提出基于事件的训练方法,包括延迟,使重量和延迟的精确梯度计算成为可能.
- 引入一种新的延迟学习算法,适用于反复出现的SNN.
- 通过学习延迟来证明SNN的性能和效率的提高.
主要方法:
- 开发了一个基于 EventProp 形式主义的基于事件的训练方法,用于有延迟的 SNN.
- 实现了一个延迟学习算法,支持每个神经元的多个峰值和反复连接.
- 评估了序列检测,阴阳,增高海德堡数字,增高语音命令和盲文读字数据集的方法.
主要成果:
- 拟议的算法成功优化了从低于最佳的初始状态延迟.
- 与没有延迟的SNN相比,分类准确性得到了提高,特别是在较小的网络中.
- 该方法显示了显著的效率增长,使用不到一半的内存,并且比最先进的延迟学习技术快26倍.
结论:
- 以学习延迟为基础的基于事件的培训是提高SNN性能和效率的有效方法.
- 对于较小的SNN架构来说,反复延迟特别有利.
- 这种方法为训练有延迟的SNN提供了一个计算效率高和节省内存的替代方案.
相关概念视频
01:35Long-term Potentiation
58.2K
Long-term potentiation, or LTP, is one of the ways by which synaptic plasticity—changes in the strength of chemical synapses—can occur in the brain. LTP is the process of synaptic strengthening that occurs over time between pre- and postsynaptic neuronal connections. The synaptic strengthening of LTP works in opposition to the synaptic weakening of long-term depression (LTD) and together are the main mechanisms that underlie learning and memory.
58.2K
01:25Long-term Potentiation
3.4K
Long-term potentiation, or LTP, is one of the ways by which synaptic plasticity—changes in the strength of chemical synapses—can occur in the brain. LTP is the process of synaptic strengthening that occurs over time between pre and postsynaptic neuronal connections. The synaptic strengthening of LTP works in opposition to the synaptic weakening of long-term depression (LTD) and together are the main mechanisms that underlie learning and memory.
Hebbian LTP
LTP can occur when...
Hebbian LTP
LTP can occur when...
3.4K
01:28Integration of Synaptic Events
3.4K
Synaptic integration mainly includes the summation of graded potentials. Graded potentials, regardless of their type, cause subtle alterations in membrane voltage, resulting in either depolarization or hyperpolarization. These incremental changes, when combined or summed, can propel the neuron toward its threshold. Consider, for example, a membrane experiencing a +15 mV shift, causing it to depolarize from -70 mV to -55 mV. In this scenario, graded potentials govern the membrane's ability to...
3.4K
01:23Propagation of Action Potentials
8.8K
The propagation of an action potential refers to the process by which a nerve impulse, or "action potential," travels along a neuron.
Neurons (nerve cells) have a resting membrane potential, with a slightly negative charge inside compared to outside. This is maintained by ion channels, such as sodium (Na+) and potassium (K+) channels, which control the flow of ions. When a stimulus, like a touch or a signal from another neuron, triggers the neuron, sodium channels open, allowing sodium ions to...
Neurons (nerve cells) have a resting membrane potential, with a slightly negative charge inside compared to outside. This is maintained by ion channels, such as sodium (Na+) and potassium (K+) channels, which control the flow of ions. When a stimulus, like a touch or a signal from another neuron, triggers the neuron, sodium channels open, allowing sodium ions to...
8.8K