Fig. 5: Results of applying the final deep learning model on a sample of random structures from the validation data set. | Microsystems & Nanoengineering

Fig. 5: Results of applying the final deep learning model on a sample of random structures from the validation data set.

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

Fig. 5

The model predicted almost all 10,000 results from the validation set exactly. The solid blue lines show the absorption curves obtained from the simulation package, and the dotted orange lines show the absorption curves predicted by the deep learning model. These curves show a one to one comparison of predicted absorption value vs real absorption value for each frequency

Back to article page