一类非线性感知器的设计与应用

Design and Application of one Non-linear Perceptron

本文设计出一类比传统感知器更加接近实际的生物感知器,根据相应的学习规则,它能够成功的执行非线性分类任务。根据突触连接的不同,如纯兴奋突触连接或等比率的兴奋与抑制突触连接,考虑三种不同的模型,得出相应的决策支持界。应用所得的学习规则,此类单层感知器也能够执行非线性的分类任务,如异或问题(XOR)的解决。

One more realistic biological version of the traditional perceptron is presented here with a learning rule which enables training of the neuron on non-linear tasks.Three different models are introduced with synaptic connections varying from the purely excitatory to a ratio of equal excitatory and inhibitory connections.Using the derived learning rule a single neuron is trained to successfully classify the XOR problem.