An Introduction to Probabilistic Spiking Neural Networks: Probabilistic Models, Learning Rules, and Applications
An Introduction to Probabilistic Spiking Neural Networks: Probabilistic Models, Learning Rules, and Applications
复制标题
DOI：
10.1109/msp.2019.2935234
复制
发表时间：
2019-11-01
影响因子：
14.9
通讯作者：
Gruening, Andre
中科院分区：
文献类型：
--
作者：
Jang, Hyeryung;Simeone, Osvaldo;Gruening, Andre
尖峰神经网络（SNN）是分布式可训练系统，其计算元件或神经元的特征在于内部模拟动力学以及数字和稀疏突触通信。突触尖峰输入的稀疏性和神经处理的对应的事件驱动性质可以通过节能硬件实现来利用，与传统的人工神经网络（ANN）相比，这可以提供显著的能量减少。然而，SNN的训练算法的设计落后于硬件实现：大多数现有的SNN训练算法都是针对生物相容性设计的，或者是通过速率编码从预训练的ANN转换而来的。
Spiking neural networks (SNNs) are distributed trainable systems whose computing elements, or neurons, are characterized by internal analog dynamics and by digital and sparse synaptic communications. The sparsity of the synaptic spiking inputs and the corresponding event-driven nature of neural processing can be leveraged by energy-efficient hardware implementations, which can offer significant energy reductions as compared to conventional artificial neural networks (ANNs). The design of training algorithms for SNNs, however, lags behind hardware implementations: most existing training algorithms for SNNs have been designed either for biological plausibility or through conversion from pretrained ANNs via rate encoding.