Recognition of Schrodinger cat state based on CNN

TL;DR

The results show that LeNet may mistakenly recognize coherent states as cat states without coherent features, while ResNet provides a feasible solution to the problem of mistakenly recognizing cat states and coherent states by traditional neural networks.

Abstract

We applied convolutional neural networks to the classification of cat states and coherent states. Initially, we generated datasets of Schrodinger cat states and coherent states from nonlinear processes and preprocessed these datasets. Subsequently, we constructed both LeNet and ResNet network architectures, adjusting parameters such as convolution kernels and strides to optimal values. We then trained both LeNet and ResNet on the training sets. The loss function values indicated that ResNet performs better in classifying cat states and coherent states. Finally, we evaluated the trained models on the test sets, achieving an accuracy of 97.5% for LeNet and 100% for ResNet. We evaluated cat states and coherent states with different α, demonstrating a certain degree of generalization capability. The results show that LeNet may mistakenly recognize coherent states as cat states without coherent features, while ResNet provides a feasible solution to the problem of mistakenly recognizing cat states and coherent states by traditional neural networks.

Figures (5)

• Figure 1: Preprocessed dataset. Randomly select 6 pictures of cat state and coherent state from a batch for display.
• Figure 2: Schematic diagram of LeNet's network structure. The model comprises four convolutional layers, two max pooling layers, two fully connected layers, and one output layer. Yellow squares denote convolutional layers, red squares denote max pooling layers, and purple squares denote fully connected layers and the output layer.
• Figure 3: Schematic diagram of ResNet's network structure. It comprises one initial convolution layer, four residual blocks, one average pooling, one full connection layer, and one output layer. Each residual block comprises two convolution layers and two batch normalization layers (not depicted in the figure), with each training cycle involving two residual blocks. Yellow squares denote convolution layers, red squares denote batch normalization layers, pink squares denote average pooling layers, and purple squares denote both the full connection layer and the output layer.
• Figure 4: Relationship between loss value and epochs. The figure shows the loss function value curve of LeNet and ResNet on the training set respectively, and the value of epoch is an integer from 0 to 100.
• Figure 5: Visualization of the LeNet predictive model. The correct prediction is displayed in blue, and the wrong prediction is displayed in red.