量子衍生神经网络模型及其在图像恢复中的应用

摘要

为提高图像恢复质量，提出一种量子衍生神经网络模型及算法。该模型为3层结构，隐层为量子神经元，输出层为普通神经元。量子神经元由量子旋转门和多位受控非门组成，利用多位受控非门中目标量子位的输出向输入端的反馈，实现对输入序列的整体记忆，利用受控非门输出中多位量子比特的纠缠，获得量子神经元的输出。基于量子计算理论设计了该模型的学习算法，该模型可从宽度和深度两方面获取输入序列的特征。仿真结果表明，该模型的图像恢复效果明显优于普通神经网络。%In order to improve the quality of image restoration , a kind of quantum-derived neural network model and algorithm are proposed .The model is composed of three layers of structure , in which , the hidden layer is made up of quantum neurons , and the output layer is made up of common neurons .The quantum neuron consists of a quan-tum rotation gate and a multi-qubit controlled not-gate.By using the information feedback of the target qubit from the output to the input end in the multi-qubit controlled not-gate, the integral memory of input sequences is real-ized .The output of the quantum neuron is obtained by the entanglement of the multi-qubit in the output of the con-trolled not-gates.On the basis of the theory on quantum computation , the learning algorithm for the model is de-signed.Through the model , the characteristics of the input sequence may be effectively obtained from two aspects including "width"and"depth".The simulation results show that the quality of image restoration of the model is ob-viously superior to that of the common neural network .