Machine learning prediction of electric field-dependent absorption coefficient in CdTe/CdS quantum dots

We investigated the electric field-induced optical absorption coefficient in CdTe/CdS core-shell quantum dots embedded within titanium dioxide (TiO₂) and silicon dioxide (SiO₂) matrices. To model these changes, we employed a comparative approach, utilizing Artificial Neural Networks (ANN), Decision Trees (DT), Random Forest Regressors (RFR), and Light Gradient Boosting Machine (LightGBM) and comparing their predictions with numerical finite element method simulations. Our findings revealed that TiO₂ embedding resulted in a redshift and amplitude increase of the absorption resonance, whereas SiO₂ embedding or isolation caused a blueshift and amplitude decrease. Notably, the Random Forest Regressor exhibited the most accurate predictions, underscoring the effectiveness of machine learning in simulating and predicting the optical properties of quantum dot systems.