Optimizing Spectral Prediction in MXene-Based Metasurfaces Through Multi-Channel Spectral Refinement and Savitzky-Golay Smoothing

2026-02-09

AI Summary

This study addresses the computational inefficiency of full-wave electromagnetic simulations for predicting the spectral response of MXene-based metasurfaces. To overcome this limitation, the authors propose an efficient deep learning framework built upon a fine-tuned MobileNetV2 architecture, enhanced with a multi-channel spectral refinement module to improve feature extraction. Additionally, Savitzky-Golay smoothing is integrated to suppress high-frequency noise in the predicted spectra. Using only a 64×64 structural input, the model rapidly and accurately predicts 102-point absorption spectra, achieving an average RMSE of 0.0245, an R² of 0.9578, and a PSNR of 32.98 dB. The method significantly outperforms baseline approaches based on standard CNNs and deformable CNNs, demonstrating both high prediction accuracy and computational efficiency.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Search and Optimization: Metareasoning and Metaheuristics
• Computer Vision: Multi-modal Vision

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

The prediction of electromagnetic spectra for MXene-based solar absorbers is a computationally intensive task, traditionally addressed using full-wave solvers. This study introduces an efficient deep learning framework incorporating transfer learning, multi-channel spectral refinement (MCSR), and Savitzky-Golay smoothing to accelerate and enhance spectral prediction accuracy. The proposed architecture leverages a pretrained MobileNetV2 model, fine-tuned to predict 102-point absorption spectra from $64\times64$ metasurface designs. Additionally, the MCSR module processes the feature map through multi-channel convolutions, enhancing feature extraction, while Savitzky-Golay smoothing mitigates high-frequency noise. Experimental evaluations demonstrate that the proposed model significantly outperforms baseline Convolutional Neural Network (CNN) and deformable CNN models, achieving an average root mean squared error (RMSE) of 0.0245, coefficient of determination \( R^2 \) of 0.9578, and peak signal-to-noise ratio (PSNR) of 32.98 dB. The proposed framework presents a scalable and computationally efficient alternative to conventional solvers, positioning it as a viable candidate for rapid spectral prediction in nanophotonic design workflows.

Problem

• spectral prediction
• MXene-based metasurfaces
• electromagnetic spectra
• solar absorbers
• computational efficiency

Innovation

• multi-channel spectral refinement
• Savitzky-Golay smoothing
• transfer learning
• MXene-based metasurfaces
• spectral prediction