基于多级神经元的神经网络及其在分类中的应用

为了提高前向神经网络的分类能力，该文将多级神经元扩展使用到多层感知器的输出层和隐含层中，并提出了量子神经网络的学习算法。通过一个实际的分类问题实验验证了该方法的有效性。实验表明，无论输出层或隐含采用多级神经元，都可以带来分类能力的提高。而当输出层采用多级神经元时，还可以导致连接的减少和训练速度的加快。

Feed Forward Neural Network Based on Multilevel Neuron and Its Application in Classification

In order to improve the classification capability of feed forward neural network, this paper extends the use of multilevel neurons for output layer and hidden layer of multiplayer perceptron network. Training algorithm for quantum neural network is proposed. The efficacy of the proposed algorithm is verified by means of simulation on a classification problem. Whatever the output layer or hidden layer uses multilevel neurons, the results show the proposed method has higher classification capability than that of traditional multiplayer perceptron network. In addition, it has been found that the neural network with multilevel output unit is often able to have fewer link weights and learn faster.