Stochastic data augmentation method for improving machine learning Data augmentation is a frequently used strategy for creating new data to improve machine learning systems for image classification, object detection ,instance segmentation, and speech recognition. Data augmentation method can solve problems of under-fitting, imbalanced data and reduce the Loss of model, therefore it can improve machine learning model. This work introduce the stochastic approach for data augmentation, such as Gaussian distribution, and Bayes inference as a method to generate new data from the existing data to create the best model.