Efficient deep learning framework for field applications
Enforcing structural similarity using deep perceptual losses extracted from the pretrained network
Self-regularized training scheme and use augmented training dataset to improve the stability of the DNN model
Model-to-model deep learning joint imaging framework using U-Net
Equivalent implementation of recurrent neural network (RNN) for simulating wave propagation using FDTD
Incorporating the forward modeling as the regularization to train the network
A Maxwell’s equations based deep learning method for time domain electromagnetic simulations
Learning the descent directions offline during the training process
