Malicious URL Classification Model Based on Hybrid Embedding and Capsule Network
As one of the most frequent cybersecurity threats,malicious URLs cause huge financial loss every year.Al-though many malicious URL detection methods have been proposed,current methods cannot make full use of the useful information provided by URL as well as extract enough discriminative features from URL,resulting in poor classification per-formance.This paper proposes a malicious URL classification model based on hybrid embedding and capsule network.The model applies highway network and capsule network to extract discriminative features from the hybrid embeddings of URLs,which improves the classification performance.The experimental results on the ISCX-URL2016 dataset prove that the pro-posed model can achieve better classification performance than other baselines,and can effectively deal with the diversity of obfuscation techniques.
万方数据
维普
