Search

The most pressing issue in network security is the establishment of an approach that is capable of detecting violations in computer systems and networks. There have been several efforts for improving it from various points of view. One example is the improvement of the classification of packets on the network, which is imperative in detecting abnormal traffic and hence any potential intrusion. Thus, this study proposes a new approach for intrusion detection that is implemented using an enhanced Bat algorithm (EBat) for training an artificial neural network. The goal of the current study is to increase the accuracy of the classification for malicious and un-malicious network traffic. The proposed study herein includes a comparison with nine other metaheuristic algorithms (conventional and new algorithms) that are used to evaluate the new approach alongside the related works. Firstly, the EBat algorithm was developed and used to select suitable weights and biases. Next, the neural network was employed using the found optimal weights and biases to realize the intrusion detection approach. Four types of intrusion detection evaluation datasets were used to compare the proposed approach against the other algorithms. The findings revealed that the proposed method outperformed the other nine classification algorithms and it is unparalleled for the network intrusion detection. [ABSTRACT FROM AUTHOR]

Arguably the most recurring issue concerning network security is building an approach that is capable of detecting intrusions into network systems. This issue has been addressed in numerous works using various approaches, of which the most popular one is to consider intrusions as anomalies with respect to the normal traffic in the network and classify network packets as either normal or abnormal. Improving the accuracy and efficiency of this classification is still an open problem to be solved. The study carried out in this article is based on a new approach for intrusion detection that is mainly implemented using the Hybrid Artificial Bee Colony algorithm (ABC) and Monarch Butterfly optimization (MBO). This approach is implemented for preparing an artificial neural system (ANN) in order to increase the precision degree of classification for malicious and non-malicious traffic in systems. The suggestion taken into consideration was to place side-by-side nine other metaheuristic algorithms that are used to evaluate the proposed approach alongside the related works. In the beginning the system is prepared in such a way that it selects the suitable biases and weights utilizing a hybrid (ABC) and (MBO). Subsequently the artificial neural network is retrained by using the information gained from the ideal weights and biases which are obtained from the hybrid algorithm (HAM) to get the intrusion detection approach able to identify new attacks. Three types of intrusion detection evaluation datasets namely KDD Cup 99, ISCX 2012, and UNSW-NB15 were used to compare and evaluate the proposed technique against the other algorithms. The experiment clearly demonstrated that the proposed technique provided significant enhancement compared to the other nine classification algorithms, and that it is more efficient with regards to network intrusion detection. [ABSTRACT FROM AUTHOR]

Intrusion detection is one of the most significant concerns in network safety. Improvements have been proposed from various perspectives. One such proposal is improving the classification of packets in a network to determine whether the classified packets contain harmful packages. In this study, we present a new approach for intrusion detection using a hybrid of the artificial bee colony (ABC) algorithm and particle swarm optimization (PSO) to develop an artificial neural network with increased precision in classifying malicious from harmless traffic in a network. Our proposed algorithm is compared with four other algorithms employed in WEKA tool, namely, radial basis function, voted perceptron, logistic regression, and multilayer perceptron. The system is first prepared, and then suitable biases and weights, for the feed-forward neural network are selected using the hybrid ABC-PSO algorithm. Then, the network is retrained using the prepared information, which has been generated from the ideal weights and biases, to establish the intrusion-detection model. The KDD Cup 1999 data set is used as the information set in comparing our proposed algorithm with the other algorithms. The experiment shows that the proposed method outperforms the four classification algorithms and is suitable for network intrusion detection. [ABSTRACT FROM AUTHOR]

This study proposes a novel approach based on multi-objective artificial bee colony (ABC) for feature selection, particularly for intrusion-detection systems. The approach is divided into two stages: generating the feature subsets of the Pareto front of non-dominated solutions in the first stage and using the hybrid ABC and particle swarm optimization (PSO) with a feed-forward neural network (FFNN) as a classifier to evaluate feature subsets in the second stage. Thus, the proposed approach consists of two stages: (1) using a new feature selection technique called multi-objective ABC feature selection to reduce the number of features of network traffic data and (2) using a new classification technique called hybrid ABC-PSO optimized FFNN to classify the output data from the previous stage, determine an intruder packet, and detect known and unknown intruders. [ABSTRACT FROM AUTHOR]

Although Hopfield neural network is one of the most commonly used neural network models for auto-association and optimization tasks, it has several limitations. For example, it is well known that Hopfield neural networks has limited stored patterns, local minimum problems, limited noise ratio, retrieve reverse value of pattern, and shifting and scaling problems. This research will propose multi-connect architecture (MCA) associative memory to improve the Hopfield neural network by modifying the net architecture, learning and convergence processes. This modification is to increase the performance of associative memory neural network by avoiding most of the Hopfield neural network limitations. In general, MCA is a single layer neural network uses auto-association tasks and working in two phases, that is learning and convergence phases. MCA was developed based on two principles. First, the smallest net size will be used rather than depending on the pattern size. Second, the learning process will be performed to the limited parts of the pattern only to avoid learning similar parts several times. The experiments performed show promising results when MCA shows high efficiency associative memory by avoiding most of the Hopfield net limitations. The results proved that the MCA net can learn and recognize unlimited patterns in varying size with acceptable percentage noise rate in comparison to the traditional Hopfield neural network. [ABSTRACT FROM AUTHOR]