Classification of Anomalous Data using SVM Classifier
Aman Mishra, Deepak Singh · 2022 International Conference on Applied Artificial Intelligence and Computing (ICAAIC) · 2022
Data security is the most challenging area for each country. Different types of attacks are introduced by many attackers and each time when an attacker tries to break the security system for anonymously accessing the sensitive data. It shows the weakness of the existing intrusion detection system but continual research efforts are taken by adding new rules to the intrusion detection system based on the recent attacks. A behavioral approach can be used to capture the attack and the attacks can be stopped through the network. IDS is a piece of software that has been traditionally used to monitor the harmful activities and generate reports. Different types of intrusion detection systems (IDS) are available nowadays, they are network-based, signature-based, and anomaly-based [1]. Many of the researchers work on anomaly-based IDS because it can prevent attacks. Anomaly-based detection techniques (ABDT) perform efficiently in fraud detection. This paper has used both Linear Support Vector Machine (LSVM) and non-linear Support Vector Machine (NLSVM) to implement a behaviora based algorithm for classifying the malicious data. For this purpose, the data will be taken from the University of New Brunswick. The data is summarized in. CSV file. This is a large dataset contains thousands data on malicious activities.