基于标签传播与重采样的不平衡数据分类方法

Unbalanced data is very common in real-life applications, and it often leads to a decrease in classification performance of classifiers. In this paper, a method for classifying imbalanced data based on label propagation and resampling is proposed. Firstly, in order to enrich the distribution of the minority class, the label propagation algorithm is used to assign pseudo labels to the test set data. Then, the samples in the test set with pseudo labels as positive examples are combined with the training set samples to form a new training set, and SMOTE-ENN is used for resampling. Finally, the classifier is trained using the resampled dataset. The experiment is conducted using 10 datasets from KEEL for validation, and the results show that our method outperforms other sampling methods in terms of AUC and G-mean indicators for classifying imbalanced data.