Machine learning-aided band gap prediction of semiconductors with low concentration doping

The fitting process of the deep learning model based on few-shot algorithm within 300 epochs is shown (a), and the best predictive model demonstrates good performance in predicting band gap values of diluted doped semiconductors (b).