【摘要】 本文在生物神经突触特性的基础上,提出了非线性神经突触神经元的概念,并以此为根据构造了一种可自学习的联想记忆神经网络模型。这种模型可以按照Hebb规则进行学习,学习机制由网络本身完成。在此模型中,由于非线性权重的引入,使此神经网络模型能以简单的结构实现网络的自学习功能。文中对网络的记忆容量和此种网络在以特定的学习方式学习后与Hopfield网络的等效性方面进行了讨论。试验表明,此种网络模型结构是有效的。更多还原

【Abstract】 The paper, based on the charactistics of biological neuron synapses, suggests the concept of non - linear neural synapse neurons, and then constructs a self - learning associative memory neural network model. This model can be used to learn in Hebb rules. The learning mechanism is performed by the neural network itself. After introducing non - linear network weights, it is easy to realize the network self - learning functions in a simple structure. The paper discusses the memory contents of the network and the equal effectivity of it compared with the Hopfield network after learning in a special learning mode. Our experiment shows this network model is feasible.更多还原