Scholarlink

High precision and efficient plasmonic sensors have paved the path for revolutionizing the biomedical industry. A one dimensional defective plasmonic grating sensor is investigated using FDTD (Finite Difference in Time Domain) method and performance of the sensor is evaluated using sensitivity. Plasmonic grating is designed on Metal-Insulator-Metal (MIM) waveguide geometry and the defect is introduced in the grating for filtering out certain frequency band. The variation of defect length versus stop band frequency is studied and the variation is used to calculate the refractive index sensitivity. The results are further evaluated using machine learning model. The model is trained using data generated by Generative Adversarial Network (GAN). The results demonstrate significant improvement in model performance, with the final model achieving an R2 Score of 0.9999. These findings underscore the importance of data augmentation in predictive modeling.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance Optimization of Plasmonic Sensor Using Machine Learning and Generative Adversial Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Priyanka Sharma,
  • Rukhsar Zafar

摘要

High precision and efficient plasmonic sensors have paved the path for revolutionizing the biomedical industry. A one dimensional defective plasmonic grating sensor is investigated using FDTD (Finite Difference in Time Domain) method and performance of the sensor is evaluated using sensitivity. Plasmonic grating is designed on Metal-Insulator-Metal (MIM) waveguide geometry and the defect is introduced in the grating for filtering out certain frequency band. The variation of defect length versus stop band frequency is studied and the variation is used to calculate the refractive index sensitivity. The results are further evaluated using machine learning model. The model is trained using data generated by Generative Adversarial Network (GAN). The results demonstrate significant improvement in model performance, with the final model achieving an R2 Score of 0.9999. These findings underscore the importance of data augmentation in predictive modeling.