Heatmap-Based Deep Learning Model for Network Attacks Classification

Heatmap-Based Deep Learning Model for Network Attacks Classification

연구 분야: Networking

학회: SN Computer Science

Networks play a vital role in our modern lives. With the emergence of expansive open networks, the security threats facing these networks have significantly escalated in recent years. We know that everything is shifting in online mode be it confidential meetings, online banking, educational meeting or whatever service we are using through the internet all are under threat and different types of intruders or attackers can attack in our network to get data from our node, so to provide the user a system in which he/she can identify that which kind of attack can take place in his system, we are making an Intrusion detection system (IDS) which will notify the user and provide the information about which kind of attack may take place. The idea is of classifying different attack in a network into various groups like dos attack, benign, SQL-injection etc. In this model we have used different types of technologies like Deep learning, Machine learning as superset and the different classification algorithms of machine learning, convolutional neural network, and heatmap as subset. We used a dataset featuring millions of rows and 46 columns each that defines different attack feature withs separate ‘label’ and ‘Attack’ columns to classify the attacks, and we will use it as training data for our model. There are models trained for the individual attacks but significantly the proposed approach will help us to classify all the potential network attacks that may occur in a single model and will provide the information about the attack using CNN.