Fig. 3
From: Supervised learning of the Jaynes–Cummings Hamiltonian
Schematic diagram of our dDNN model. A D-UNet receives noisy input and outputs the denoised data. A separately trained FNN takes the resulting output and predicts the sought-after parameters. In the D-UNet diagram, \(\text {D}_1\), \(\text {D}_2\), \(\text {D}_3\), and N represent the dimensions of each layer’s output. The green trapezoids inside the D-UNet indicate that the number of nodes in each layer increases toward the model’s center. A more detailed structure of the D-UNet module is shown in the inset, indicating how the layers are connected. A skip connection, denoted by a dotted arrow, indicates that the layer is connected to a non-neighboring layer.
