Deep Learning With Plasma Plume Image Sequences For Anomaly Detection And Prediction Of Growth Kinetics During Pulsed Laser Deposition

This article discusses the use of deep learning techniques in materials synthesis with physical vapor deposition techniques. The study uses a small dataset of 127 samples and employs data augmentation to improve model generalization. The article also covers the use of the mean square error loss function and the Adam optimizer, as well as hyperparameter tuning using Ray Tune and the Optuna algorithm. The authors found that training with data augmentation improved model predictions.