Public Datasets (6 public datasets used for evaluation)

该数据集用于评估DTOR方法在解释不同异常检测模型检测到的异常方面的有效性。这些数据集包含了异常数据和正常数据点，其中训练集中异常的比例是预定义的（对于高斯混合模型和隔离森林为5%，对于支持向量机约为50%）。这些数据集的规模多样，测试集每组包含50个样本，任务旨在进行异常检测及其解释。

This dataset is designed to evaluate the effectiveness of the DTOR method in explaining anomalies detected by various anomaly detection models. This dataset collection includes both anomalous and normal data points, where the proportion of anomalies in the training set is pre-defined: 5% for Gaussian Mixture Models (GMM) and Isolation Forests, and approximately 50% for Support Vector Machines (SVM). The datasets have varied scales, with each test set containing 50 samples. The task focuses on both anomaly detection and the explanation of the detected anomalies.