Tabular Anomaly Detection Datasets

该数据集包含了五个公开可获得且广泛使用的表格数据集，旨在评估adVAE模型在异常检测任务中的表现。每个数据集都被划分为80%的正常数据用于训练，以及20%的正常数据加上所有的异常值用于测试。评估这些数据集时，我们采用了精确度、召回率、F1分数、AUC（曲线下面积）以及平均精度等指标。

This dataset contains five publicly available and widely utilized tabular datasets, which are designed to evaluate the performance of the adVAE model in anomaly detection tasks. Each dataset is partitioned into 80% normal data for training, while the remaining 20% normal data plus all outliers are reserved for testing. When evaluating these datasets, we adopt metrics including precision, recall, F1-score, AUC (Area Under the Curve), and average precision.