摘要
研究网络异常入侵检测问题。将SVM和融合技术应用于入侵检测领域,解决了传统SVM算法易产生训练参数选择不当,检测效率和分类精度低的问题。实现了对特征库中各特征量根据报警信息时间序列的预测进行优化和更新,有效地降低了算法的时间复杂度和空间复杂度,提高入侵检测系统对已有特征量对应攻击的识别效率。实验结果表明,该融合算法训练时间短、分类精度高、测试时间减少,误报率和漏报率低,有效提高了入侵检测系统的准确性和实时性。是一种有效可行的入侵检测方法。
Research network anomaly intrusion detection problem.The SVM and fusion technology applied in intrusion detection field,solve the traditional SVM algorithm is easy to produce training parameter improper selection,detection efficiency and the problem of low accuracy of classification.Realize the characteristics of each characteristic library according to the alarm information time series prediction optimization and update,effectively reduces the time complexity of the algorithm and spatial complexity,enhance the intrusion detection system to have the characteristic parameter corresponding to the recognition efficiency of the attack.The experimental results show that the fusion algorithm training time is short,the classification accuracy is high,the test time,reduce false positives and fail to low,effectively improve the accuracy of the intrusion detection system and real time.Is a kind of effective and feasible intrusion detection method.
出处
《科技通报》
北大核心
2013年第5期167-172,共6页
Bulletin of Science and Technology
基金
河南省科自然科学项(112102210335)