I constructed a residual neural network. The model input image and output image. When the model is used, it is an autoregressive method. However, my model generalization ability is very poor. Is there any way to improve the model generalization ability ? Or how to change the structure of the neural network ? I only show a small part of the data here, and I have 50000 iterations. The generalization ability of the results is still very poor, and the model has the problem of over-fitting.
