BROWSE
Detail
Conference Paper
Training-Dataset-Free Artificial Neural Network Approach for Inverse Design of Multilayered Nanophotonic Films
Cited - time in 
Share 
Share
- Authors
- Jun Hee Han, Kyu Sung Lee
- Issue Date
- 2026-04
- Citation
- Materials Research Society (MRS) Meeting 2026 (Spring), pp.1-1
- Publisher
- Materials Research Society
- Language
- English
- Type
- Conference Paper
- Abstract
- This study presents a novel neural network architecture that embeds physical laws directly into its nodes and connections This enables efficient calculation of gradients for each input parameter during forward and backward propagation, bringing the computational benefits of the adjoint method into the neural network framework The model is designed for multilayered nano thin film structures, reflecting their stacked layer characteristics through a recurrent network topology This structure allows the seamless addition of new layers without rebuilding the network, enabling design for desired optical performance.