KDD CUP 99

该数据集是DARPA98数据集的过滤版本，主要用于网络入侵检测的基准测试。它包含5种模式类别：正常、拒绝服务（DoS）、未授权的远程访问（U2R）、远程到本地访问（R2L）和探测（Probe）。数据分布不均，正常和DoS类别的数据占总量的98.61%。在规模上，训练集包含4,898,431个数据点（其中1,074,992个是唯一的），测试集包含2,984,154个模式（其中311,029个是唯一的）。该数据集的任务是网络流量中的异常检测。

This dataset is a filtered version of the DARPA98 dataset, primarily used as a benchmark for network intrusion detection. It contains five pattern categories: Normal, Denial of Service (DoS), Unauthorized Remote Access (U2R), Remote to Local Access (R2L), and Probe. The data is unevenly distributed, with the Normal and DoS categories accounting for 98.61% of the total dataset. In terms of scale, the training set includes 4,898,431 data points, of which 1,074,992 are unique; the test set contains 2,984,154 patterns, with 311,029 of them being unique. The task of this dataset is anomaly detection in network traffic.