中文摘要：

针对数据非平稳分布而导致的异常检测差错率较高的问题，提出了一种可用单类1／4球体支持向量机模型参数表示的异常率参数动态调整自适应算法。该算法在线迭代运行，依次生成基于训练集的最优模型，最后提出一种可以迭代得到训练集异常率的参数确定算法，可发现数据集中的异常。仿真实验结果表明，与异常率固定的静态模型相比，该算法可以利用训练集生成最优模型，进而实现分类集差错率最小。

英文摘要：

Aiming at the problem of the high error rate of anomaly detection caused by the non-stationary distributions in the data, this paper proposed an adaptive algorithm that could dynamically adjust the anofnaly rate parameter, which could be re- presented by a model parameter of a one-class quarter-sphere support vector machine. This algorithm operated in an online, it- erative manner producing an optimal model for a training set, which was presented sequentially. Finally, it introduced an algo- rithm that was able to iterate to the anomaly rate in the training set to discovery the anomaly in the training set. Simulation re- suits demonstrate that this algorithm is capable of constructing optimal models for a training set that minimizes the error rate on the classification set compared to a static model, where the anomaly rate is kept stationary.