Predicting optical properties of different photonic crystal fibers from 2D structural images using convolutional neural network and transfer learning

Machine Learning Assisted Enhancement in a Two-Dimensional Material’s Sensing Performance

“…DL methods like CNNs have shown impressive performance in managing augmented materials, as they can discover complex links and patterns in data, increasing sensing efficiency. , Generative models like GANs can be used to produce realistic synthetic data, particularly in image-based sensing jobs . Pretrained models can be refined on specific sensing tasks using transfer learning approaches. , DL and traditional machine learning algorithms can enhance 2D material sensing performance by automatically extracting features and capturing intricate patterns, while traditional methods rely on manual feature engineering and struggle with complex data relationships. Experiment design and data augmentation strategies are essential for maximizing the strengths of these algorithms.…”

“…148 Pretrained models can be refined on specific sensing tasks using transfer learning approaches. 160,161 DL and traditional machine learning algorithms can enhance 2D material sensing performance by automatically extracting features and capturing intricate patterns, while traditional methods rely on manual feature engineering and struggle with complex data relationships. Experiment design and data augmentation strategies are essential for maximizing the strengths of these algorithms.…”