Deep Learning Enabled Inverse Design of Interlayer-Coupled Chiral Metamaterials

Machine learning has shown significant potential in metamaterial design. The application of deep learning for chiral metamaterials, especially in interlayer coupling effects, is still challengeable. Here, we introduce a novel approach using deep neural networks to learn the feature mappings between two chiral characteristics (circular dichroism, CD, circular conversion dichroism, CCD) and their geometric parameters, resulting in a reverse design neural network. The findings present a fresh perspective on designing interlayer-coupled chiral metamaterials, which can be applied to various domains including terahertz multifunctional devices and sensing.