Deep-Learning-Based Source Reconstruction Method Using Deep Convolutional Conditional Generative Adversarial Network

This article proposes a novel deep-learning (DL)-based source reconstruction method (SRM). The proposed DL-based SRM employs the deep convolutional conditional generative adversarial network (DCCGAN), which only demands one-transmitter single-frequency far-field measurement on electromagnetic (EM) scattered field as its input and further predicts the equivalent source on target scatterers. The proposed DCCGAN includes the generator (G) with an EM forward simulator and the corresponding discriminator (D), both consisting of the complex-valued deep convolutional neural networks (DConvNets). During the offline training, the generator learns the distribution between the measured scattered field data and the corresponding equivalent source on target scatterers, while the discriminator determines whether the presented equivalent sources are real or fake. Therefore, the proposed DCCGAN can generate the unknown equivalent source from measured scattered field data, by learning the distribution between the known equivalent sources and their corresponding field. Furthermore, the proposed DL-based SRM can overcome the limitation of conventional methods, involving high computational cost and strong ill-conditions. Consequently, the proposed DL-based SRM can realize the reconstruction of the equivalent source with higher accuracy and lower computation complexity. Numerical examples have demonstrated the feasibility of the proposed DL-based SRM, which opens the new path for DL-based EM computation approaches.