Fig. 22: Description of physics-in-network phase recovery. | Light: Science & Applications

Fig. 22: Description of physics-in-network phase recovery.

From: On the use of deep learning for phase recovery

Fig. 22

a A physics-based iterative algorithm. b A corresponding unrolled neural network. The iteration step \(h\) with algorithm parameters \(\omega\) in (a) is unrolled and transferred to the network layers \({h}_{1}\), \({h}_{2}\),…, \({h}_{n}\) with network parameters \({\omega }_{1}\), \({\omega }_{2}\),…, \({\omega }_{n}\) in (b). The unrolled neural network is trained with the dataset in an end-to-end manner

Back to article page