论文部分内容阅读
提出一种量子神经网络模型及算法.该模型为一组量子门线路.输入信息用量子位表示,经量子旋转门进行相位旋转后作为控制位,控制隐层量子位的翻转;隐层量子位经量子旋转门进行相位旋转后作为控制位,控制输出层量子位的翻转.以输出层量子位中激发态的概率幅作为网络输出,基于梯度下降法构造了该模型的学习算法.仿真结果表明,该模型及算法在收敛能力和鲁棒性方面均优于普通BP网络.
A quantum neural network model and algorithm is proposed. The model is a set of quantum gate lines. The input information is represented by quantum bits, which are controlled by the quantum rotation gate to control the inversion of hidden layer qubits. After the phase rotation of the quantum revolving door is used as a control bit, the quantum bit inversion of the output layer is controlled. The probability amplitude of the excited state in the output layer quantum bit is taken as the network output and the learning algorithm of the model is constructed based on the gradient descent method. , The model and algorithm are better than ordinary BP network in convergence ability and robustness.