Paper Title
Hyperspectral Images Classification with Deep Bayesian Neural Networks

Abstract
The classification of Hyperspectral Image HSI is important in various fields where more discriminative characteristics are provided by the hundreds of narrow-band radiation information. Bayesian CNN is a very powerful method used in various difficult classification problems. In this work, Bayesian CNN model was built and applied on Pavia dataset. In order to compare the Bayesian results with other methods, two other approaches were applied. Machine learning methods are more frequently used, which uses both labeled and unlabeled data to fit the model. SVM and RF were used. Pretrained deep learning models were also applied. The results show that the Bayesian CNN method gives the best accuracy of 99%. Among pretrained deep learning networks, Xception gives the best 97%. SVM with the Radial Basis Function (RBF) kernel gives 96% accuracy. Keywords - Hyperspectral Image, Pavia University Dataset, SVM, RF, Bayesian CNN, Pretrained Deep Learning Model, XCEPTION.