关键词: 自动控制技术, 异常检测, 支持向量数据描述, 支持向量机, 最小闭包球

Abstract: To efficiently resolve outlier detection problem in large scale data sets, an efficient outlier detection algorithm based on Support Vector Data Description (SVDD) was proposed. The key concept of the algorithm is as follows: First, the spherical boundary containing oneclass dataset is obtained via SVDD; then the test datasets are classified with this boundary. In addition, the minimum enclosing ball algorithm is applied to optimize the SVDD detection classifier. The algorithm is compared with other outlier detection algorithms by experiments with benchmark dataset from UCI data set repository and intrusion detection data sets. Results show that the proposed algorithm consumes less running time meanwhile achieves higher detection accuracy.

Key words: automatic control technology, outlier detection , support vector data description, support vector machine, minimum enclosing ball