In the realm of cybersecurity, the detection of Distributed Denial of Service (DDoS) attacks remains a critical challenge, exacerbated by the emergence of unknown attack patterns. This study proposes a novel approach termed Reciprocal Points Learning for Open-Set Recognition in DDoS attack detection. Leveraging Passive Aggressive, Random Forest, and Decision Tree algorithms, the methodology focuses on features extracted from network traffic data including flow duration, packet characteristics, and statistical flow metrics. The system aims to differentiate between normal network behavior and both known and unknown DDoS attacks, enhancing the robustness of detection mechanisms in dynamic and evolving threat landscapes. Experimental evaluation on a comprehensive dataset underscores the efficacy of the proposed framework in achieving high accuracy and reliability, thus contributing to the advancement of proactive cybersecurity strategies against sophisticated network attacks.

Keywords: Passive Aggressive, Random Forest and Decision Tree Classifier.