Figure 2 | Scientific Reports

Figure 2

From: A cyclical deep learning based framework for simultaneous inverse and forward design of nanophotonic metasurfaces

Figure 2

Schematic illustration of (a) SNN model architecture composed of CNNs and FC-NNs. It takes 2D cross sectional structural design image as input and predicts it’s optical response and (b) cGAN model architecture to generate structural designs. The complete architecture consists of two networks: a generator (G) and a discriminator (D). G accepts optical response and a random noise distribution z to generate structural design. D evaluates real structural designs from generated structural designs as real or fake.

Back to article page