Confusion Matrices

The one-vs-rest approach considering the multiple-label classification and micro-averaging calculations were followed for result interpretations of TP, TN, FP, and FN.针对多标签分类任务，本研究采用一对多（one-vs-rest）方法，并结合微平均（micro-averaging）计算规则，对真正例（True Positive, TP）、真反例（True Negative, TN）、假正例（False Positive, FP）以及假反例（False Negative, FN）的结果进行解读。