No abstract

No abstract

No abstract

The rate of internet traffic in the digital world has expanded fast due to technological improvement. Because of the large number of internet users, there is a large volume of network traffic, and it is the most common and challenging task to evaluate, process, and store the data. The intrusion detection systems enhance the performance of early attacks prediction and malicious attack detection. The exiting techniques met a few challenges in terms of feature selection, computational cost, accuracy, higher dimensionality, computational time and so forth. To solve these issues, we proposed a novel deep learning model for intrusion detection in big data. The data was collected from NSL‐KDD, KDD‐Cup99, and UNSW‐NB15 datasets. Initially, the big data framework with a testbed is set up for big data analytics and processing in which the required information from a large volume is extracted. The datasets are stored in the Hadoop Distributed File System, which reduces the number of latencies in the distributed process. Finally, the intrusion detection from big data is performed using convolutional neural network‐based Hybrid Whale Tabu Optimization algorithm and it effectively classifies whether the data is intrusion is non‐intrusion. The proposed method provided appropriate and superior experimental results using the state‐of‐art method with various performance measures such as accuracy, Matthews correlation coefficient, balanced accuracy, specificity, sensitivity, precision, and F‐measures.

No abstract