The focus of industry experts and academicians on developing new deep learning (DL) algorithms is intensifying. DL algorithms show greater performance and accuracy based on conditions like input data, hyperparameters, data size, and parameters. This paper presents the developmental changes that have taken place in the DL method and algorithms and use in the direction of intelligent diagnosis for rotating machinery faults. The transformations of DL algorithms’ architecture variables to perform intelligent fault diagnosis are compared. Further characterization of DL methods and algorithms for various types of rotating machinery faults like rolling element bearing misalignment imbalance and looseness is compared.