You have attempted to access the full-text of an Early Posting article. Access is available via an institutional subscription.
See the Early Posting FAQ page for additional information.
See the Early Posting FAQ page for additional information.
Cross-Domain Heterogeneous Metasurface Inverse Design Based on Transfer-Learning Method
Optics Letters- received 11/29/2023; accepted 04/11/2024; posted 04/12/2024; Doc. ID 514212
- Abstract: In this paper, a transfer learning method is proposed to complete design tasks on heterogeneous metasurface datasets with distinct functionalities. Through fine-tuning the inverse design network and freezing the parameters of hidden layers, we successfully transfer the metasurface inverse design knowledge from the electromagnetic-induced transparency (EIT) domain to three target domains of EIT (different design), absorption, and phase-controlled metasurface. Remarkably, only 5% of the target domain samples are required to complete the training process in comparison to the source domain datasets. This work presents a significant solution to lower the data threshold for inverse design process and provides the possibility of knowledge transfer between different domain metasurface datasets.