ML Based Web Vulnerability Detection

Vulnerability detection is easy to come by these days, thanks in part to the last decade which has shown signature based intrusion detection systems are not capable of detecting new and difficult cyber-attacks. In such an environment, network security is becoming a hard problem with the proliferation of applications and computer networks. This article introduces web vulnerability detection using KDDCup’99 dataset which is the standard benchmark for intrusion detection. The paper presents the techniques to detect spam and categorize it as vulnerable or non-vulnerable data. To address this, we propose a hybrid model to combine Random Forest with LSTM in order to enhance detection rate and efficiency. The complete model with data pre-processing, feature extraction and classification is provided by the framework for the identification regime. This study also suggests the prospect of combining conventional machine learning method and deep learning to increase security for web application system beyond cyber environment.