Experiments with Neural Networks Training Algorithms for DDoS attacks Detection
Attacks on servers to bring their services down is growing rapidly. There are several possible methods through which attackers can attack a server. Distributed denials of service (DDoS) attacks are one of these attacks types in computing environments. In this paper, we discuss the possible ways of DDoS attacks detection using artificial neural network based techniques. Here a feed forward neural network is designed to detect DDoS attacks in a network based environment. Performance of feed forward neural network is analyzed by applying three different training algorithms. These algorithms are (1) Scaled Conjugate Gradient (2) Bayesian Regularization and (3) Gradient Descent. Results show that Bayesian Regularization algorithm is better among these three training algorithms and it has less number of misclassifications.