Abstract: The deep learning-based fault diagnosis method faces critical challenges due to the scarcity of fault samples. This article proposes a regularized stable convolutional generative adversarial ...