講演情報
[9p-PB6-20]A Physics-Guided Machine Learning Framework for Two-Photon Absorption Spectra Determination
〇(M2C)Ian Ribeiro Andrade1, Leandro H. Zucolotto Cocca1 (1.Fed. Univ. of Goias)
キーワード：
Imidazos、Neural Network、Two Photon Absorption
This study combines Principal Component Analysis (PCA) and a Physics-Guided Neural Network (PGNN) to predict the non-linear optical properties of imidazo derivatives. A PGNN was engineered to predict two-photon absorption (2PA) cross-sections. Extracting parameters from linear inputs, the network analytically computes 2PA spectra via the Sum-Over-States (SOS) formulation. Incorporating a 10% data augmentation to simulate experimental noise enhanced the model's robustness. Validated on novel molecules, it successfully recovered the 2PA spectral envelopes. This data-driven approach provides a highly efficient computational tool to accelerate the design of new non-linear photonic materials.