Experimental Dataset and Metasurface Design Files for "All-wave computational ultrasonic fingerprint identification with metasurface-driven loop-diffractive neural network"

This repository contains the experimental dataset and metasurface design files used in the study: “All-wave computational ultrasonic fingerprint identification with metasurface-driven loop-diffractive neural network”. The dataset supports the experimental validation of a loop-diffractive neural network (loop-DNN) architecture for ultrasonic fingerprint identification. Contents of this repository include: 1. LDNN_Experiment_and_Figure_Data.xlsx Experimental data used for generating the figures and analysis presented in the manuscript. 2. loop-DNN_learned_metasurface.STL STL file of the optimized diffractive metasurface designed using the trained loop-DNN model. The metasurface design was obtained through a physics-informed training procedure that jointly optimizes phase, amplitude, and switch-state modulation of meta-atoms. These data files support the experimental results reported in the manuscript and enable reproducibility of the metasurface-based wave computing framework.本仓库收录了研究论文《基于超表面（metasurface）驱动环形衍射神经网络的全波计算超声指纹识别》（All-wave computational ultrasonic fingerprint identification with metasurface-driven loop-diffractive neural network）中使用的实验数据集与超表面设计文件。 本数据集用于支撑针对超声指纹识别任务的环形衍射神经网络（loop-diffractive neural network，简称loop-DNN）架构的实验验证。 本仓库的内容如下： 1. LDNN_Experiment_and_Figure_Data.xlsx：用于生成论文所载图表及开展分析的实验数据集。 2. loop-DNN_learned_metasurface.STL：基于经训练的loop-DNN模型所设计的优化衍射超表面的STL文件。 该超表面的设计方案通过物理信息驱动训练流程获得，该流程可联合优化超原子（meta-atoms）的相位、振幅与开关状态调制特性。 上述数据集文件可支撑论文报道的实验结果验证，并实现基于超表面的波计算框架的研究可复现性。