Domain name server (DNS) filtering service using Threat intelligence feeds and AIML Techniques

Cybersecurity threats constantly evolve, necessitating robust detection mechanisms. This study proposes a twofactor approach combining signaturebased and anomalybased detection systems for identifying packetbased attacks. Leveraging machine learning algorithms—Decision Trees, Random Forest, and GaussianNB—the system aims to enhance detection accuracy and efficiency. Signaturebased detection relies on known patterns, while anomalybased detection identifies deviations from established norms. The study evaluates the performance of these algorithms in detecting attacks on network packets. Through comprehensive experimentation and analysis, this research assesses the efficacy of each algorithm in accurately pinpointing various types of attacks. Results showcase the strengths and limitations of each model in detecting sophisticated attacks. The findings underscore the importance, harnessing the strengths of both signature and anomaly detection, and demonstrate the potential of machine learning algorithms in fortifying network security against evolving cyber threats. 

Keywords: Decision Tree, Random Forest, GaussianNB. MLP Classifier