基于量子门组的量子神经网络模型及其应用

摘要

In order to improve the performance of quantum neural network, considering the current research of neural net-work mechanism, quantum neuron model based on quantum gate set is proposed and Quantum Gate Set Neural Network (QGSNN)is established. The structure of QGSNN contains input layer, hidden layer and output layer. The input is quan-tum training samples after transformed. Quantum rotation gate and universal quantum gate are used for rotation selection overturn and aggregation, and the network parameters are updated. The results after trained are output. The generalization performance of QGSNN is proved in mathematics, and the proposed method is verified by two simulation experiments. The results indicate that, compared with common neural network and common quantum neural network, QGSNN has bet-ter effect on generalization performance, robustness accuracy and execution time.%为进一步提高量子神经网络的性能，结合目前神经网络机理的研究进展，提出了一种基于量子门组的量子神经元模型，建立了量子门组量子神经网络（Quantum Gate Set Neural Network，QGSNN）。该算法由输入层、隐含层和输出层组成，该算法将转换后的量子态训练样本作为输入。利用量子旋转门和通用量子门完成旋转、选择、翻转和聚合等一系列操作，并完成了网络参数的更新。将训练后的结果输出。QGSNN算法的泛化能力在数学上得到了证明，并利用两个仿真实验对该方法进行验证。实验结果表明，与普通神经网络和普通量子神经网络相比，QG-SNN算法在泛化性能、鲁棒性、准确率和执行时间等方面具有较好的效果。