Fig. 6: Neural network model training and computational efficiency. | Microsystems & Nanoengineering

Fig. 6: Neural network model training and computational efficiency.

From: Finding the optical properties of plasmonic structures by image processing using a combination of convolutional neural networks and recurrent neural networks

Fig. 6

a Training loss and testing loss as the model trains. The code saves the model with the lowest loss on the test data set as the best model. Both of the y axis curves are shown on a logarithmic scale. b A comparison of computation time between simulation and deep learning on 20 different structures

Back to article page