A Survey of Comparative Learning Methods
Gang Fan, Zhang Ya, Bo Li, Yixing Li, Xue Wang, Furong Wang · 2023
Recently, contrast learning has become an important approach in unsupervised learning, which is very good at classifying and predicting unlabeled data, and the best contrast learning models in recent years have been able to achieve the prediction accuracy of supervised learning models. This paper analyzes the algorithmic ideas of contrast learning, discusses the typical contrast learning models in recent years, analyzes their model structures, compares the features and differences among the models, and looks forward to the future directions of contrast learning methods that need to be improved, providing valuable references for subsequent engineering application research.