Related Experiment Video
Updated: May 5, 2026
05:57
Characterization of SiN Integrated Optical Phased Arrays on a Wafer-Scale Test Station
Published on: April 1, 2020
8.0K
Hybrid Frameworks Integrating Deep Learning and Optimization Methods for Inverse Design in Nanophotonics
Harit Keawmuang1, Shiqi Hu1, Trevon Badloe2,3
1Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea.
ACS Applied Materials & Interfaces
|May 30, 2025
Summary
Artificial intelligence (AI) is revolutionizing nanophotonics inverse design through hybrid frameworks. These models combine deep learning with classical optimization for efficient, fabrication-feasible nanoscale device designs.
Area of Science:
- Nanophotonics
- Computational Science
Background:
- Artificial intelligence (AI) is a transformative tool in nanophotonics.
- Inverse design of nanoscale devices is a key area of development.
Purpose of the Study:
- To explore the trend of AI-driven approaches in nanophotonics inverse design.
- To focus on hybrid frameworks combining deep learning and classical optimization.
Main Methods:
- Hybrid models integrating deep learning (e.g., neural networks) with classical optimization (e.g., adjoint methods, evolutionary algorithms).
- Physics-informed neural networks (PINNs) embedding physical laws into the learning process.
Main Results:
- Hybrid frameworks offer faster convergence and higher design efficiency compared to standalone methods.
- These approaches enable exploration of diverse, fabrication-feasible solutions for nanophotonic devices.
- PINNs reduce data dependency and enhance interpretability.
Conclusions:
- AI-driven hybrid frameworks are advancing nanophotonics inverse design.
- These methods facilitate scalable and practical innovations in nanophotonic technologies and functional material engineering.
