针对入侵检测的效率问题,提出一种将感知哈希用于最邻近分类的入侵检测方法,并提出建立杂交带集合来提高入侵检测速度.具体做法是先将训练样本转换为一段感知哈希摘要,再由相同的哈希摘要的训练样本构成的集合按照一定的规则建立杂交带集合;入侵检测阶段计算被检测对象的感知哈希摘要,并根据被检测对象的哈希摘要找到其在训练集合上对应的子集合从而完成粗粒度分类,然后在该子集合对应的杂交带上根据投票选择完成细粒度分类.实验结果表明该方法不仅保证了检测的精度还提高了入侵检测速度.%In order to solve the problem of intrusion detection efficiency,this paper proposed an intrusion detection method using perceptual hash for nearest neighbor classification,and proposed to establish a collection of hybridization bands to improve intrusion detection speed.In practice,the training sample was first transformed into a perceptual hash summary,and then the collection of hybridization bands was established according to certain rules on the collection formed by the training samples of the same hash abstract.The intrusion detection phase calculated the perceptual hash digest of the object to be detected and found the corresponding subset on the training set according to the hash digest of the object to be detected to finish the coarse-grained classification.Then the sub-collection of the corresponding hybridization band voted based on the completion of fine-grained classification.Experimental results showed that this method not only guaranteed the accuracy of detection but also improved the speed of intrusion detection.
展开▼
