Abstract: The safety as well as reliability of computers networks and systems, essential modern infrastructures, are now urgently threatened by DDoS attacks. DDoS attack detection must be accomplished even before prevention strategies can be implemented. However, because to a flaw in ML/DL-based systems known as the Open Set Recognition (OSR) problem, full-scale success is still out of reach. An ML/DL-based system handling new instances that arent taken from either the distribution system of a training data in this situation. It occurred when comes to DDoS attack detection because these attacks technology is always changing, traffic patterns are shifting.