machine-learning - 尖峰神经网络

Question

已结束。此问题正在寻求书籍、工具、软件库等的推荐。它不满足 Stack Overflow guidelines 。目前不接受答案。

---

我们不允许提出寻求书籍、工具、软件库等推荐的问题。您可以编辑问题，以便用事实和引文来回答。

去年关闭。

Improve this question

在尖峰神经网络领域，应该从哪本书开始？我知道郭士纳的"Spiking Neuron Models" ，出版于 2002 年。有没有更新的书，或者更合适的书？我有数学和人工神经网络背景。

如果该领域有一些好的文章或概述，也将它们添加到列表中。

谢谢。

稍后编辑

卡雷尔的回答:

" It depends what do you mean by spiking neural networks - there are at least several basic points of view. Gerstner represents the first one - he is focused on modelling of biological neurons. And his book from 2002 is really good starting point for understanding bio-physical models of neuron. It the past it was possible to find this book also in html ..

On the other hand by ¨Spiking neuron" in the computer science context is usually meant the SRMo model (Spike Response Model), which can be used also as an alternative to classical percepron-based networks.

This model is described very well in the works of Wolfgang Maass (http://www.igi.tugraz.at/maass/). He has focused on the computational power of the model and he compares the SRM model with percepron and RBF-unit.

If you want to use the model in a network I recommend to you works of Sander Bohte (http://homepages.cwi.nl/~sbohte/) who derived SpikeProp algorithm.

(I personally derived a variant of SpikeProp which was fast enough to be used for real-word applications.) "

Answer 1

尖峰神经网络 (SNN) 或脉冲神经网络 (PNN) 是更接近地模拟自然神经机制行为的人工神经网络 (ANN)。我想向您推荐以下基础书籍:

• Thomas Trappenberg 的《计算神经科学基础》
• 理论神经科学:神经系统的计算和数学建模，作者:Peter Dayan
• 脉冲神经元模型:单神经元、群体、可塑性，作者:Wulfram Gerstner

我个人推导了远程监督方法(ReSuMe)的一个变体，与 Filip Pononulak 引入的 ReSuMe 相比，它具有更好的学习率和形态学优势。

同时，我想列出一些处理 SNN 的模拟器工具。我玩过的大多数游戏都是基于 Python 的，所以也请考虑到这一点。可能还有更多基于其他语言的其他内容。

• 无政府状态
• Brian2
• 能哥
• Elephant
• 神经元
• PyNN
• NEST / pynest
• PCSIM
• Pypcsim
• BindsNet