Efficient Inverse Extreme Learning Machine for Parametric Design of Metasurfaces
Li‐Ye Xiao, Fu-Long Jin, Bing‐Zhong Wang, Qing Liu, Wei Shao
Abstract
To make metasurface design faster and more convenient, a new inverse extreme learning machine (ELM) is proposed in this letter. Compared with the traditional forward model based on artificial neural networks, the proposed inverse ELM uses a small number of training samples to meet the modeling standard, and it directly obtains the satisfying geometry without an optimization process. Meanwhile, to enhance the modeling performance, the data classification technique and a cascade of two ELMs are contained in the tailored scheme, where the second ELM is applied to error correction. A numerical example of an interdigital resonator metasurface is employed to verify the effectiveness and efficiency of the proposed model.