摘要： 根据级联结构的特征，针对入侵检测问题改进AdaBoost算法。改进的AdaBoost算法对参数求解方法、初始权值和判决阈值都进行调整，使弱分类器的加权参数不但与错误率有关，还与其对异常样本的识别能力有关。该算法能够有效地降低分类器的误警率，使其更适用于入侵检测，仿真实验证明了该算法的有效性。

关键词: 入侵检测, 改进的AdaBoost, 级联结构

Abstract: This paper proposes an improved AdaBoost algorithm aiming at intrusion detection according to the features of cascade structure. The improved AdaBoost adjusts the method of acquire parameters, the initial weights and decision threshold so that the weighted parameters of weak classifiers are determined by not only the error rates, but also their abilities to recognize the abnormal samples. The algorithm can decrease the false alarm ratio of classifiers, so it is more adaptive to the intrusion detection. The validity of the improved AdaBoost algorithm is proved by the following simulation experiment.

Key words: intrusion detection, improved AdaBoost, cascade structure

中图分类号: 

• TP393

• TP181