Artificial Intelligence-Based Inverse Design of Plasmonic Structures: A Comparative Review of Conventional and Data-Driven Methods

This study reviews traditional simulation-based methods and artificial intelligence (AI) approaches for the inverse design of plasmonic structures. Conventional techniques such as the Finite Element Method (FEM), Finite-Difference Time-Domain (FDTD), and Beam Propagation Method (BPM) provide accurate electromagnetic predictions but are computationally demanding, especially in large multi-dimensional design spaces. AI-driven approaches, including machine learning, deep learning, and generative models like Generative Adversarial Networks (GANs) and Conditional Tabular GANs (CTGANs), offer faster predictions of structural parameters from optical targets and enable synthetic dataset generation to address data scarcity. The analysis outlines the strengths and limitations of both strategies, emphasizing their complementary role in advancing high-performance plasmonic devices. Particular focus is placed on their importance for sixth-generation (6G) communication systems, which require high-speed, energy-efficient, and densely integrated optical hardware