Intrusion Detection System on Big data using Deep Learning Techniques
Priyanka Dahiya1, Devesh Kumar Srivastava2

1Priyanka Dahiya*, Manipal University Jaipur, Rajasthan. Department of School of Computing-CA, DIT University, Dehradun.
2Devesh Kumar Srivastva, Department of School of Computing & Information Technology, Manipal University Jaipur, Rajasthan, India.
Manuscript received on January 12, 2020. | Revised Manuscript received on January 22, 2020. | Manuscript published on February 10, 2020. | PP: 3242-3247 | Volume-9 Issue-4, February 2020. | Retrieval Number: D2011029420/2020©BEIESP | DOI: 10.35940/ijitee.D2011.029420
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: Big data is the huge amount of data with different types of V’s: Velocity, Variety as well as Volume. It can be semi-structured, unstructured or structured, due to which it is not easy to analyze the data. To extract the hidden knowledge and to detect the attacks on large amount of data new architecture, techniques, algorithms, and analytics are required. Using traditional techniques to detect attacks is very difficult. In this paper, the detailed review has been done on intrusion detection on various fields using deep learning and gives an idea of applications of deep learning. The number of attacks has been increased in computer networks. A powerful Intrusion Detection System (IDS) is required to ensure the security of a network. Based on review, it is found that some studies have been done in this field, but a deep and exhaustive work has still not been done. Many researchers proposed an IDS using deep learning for unforeseen and unpredictable attacks but not for Big Data. The proposed work is based on Deep learning based intrusion detection System for big datasets named hybrid-DeepResNet-RNN run till 1,000 epochs with learning rate varying range [0.01-0.5] and three ensemble techniques, Random Forest, Decision tree regression and Gradient Boosting Tree (GBT). It is used to develop the hybrid, secure, scalable NIDS which is based on deep learning and big data techniques. The proposed classifiers produce a more reliable classification than a single classifier. The experimental results are in terms of detection rate (98.86%), false positive rate (1.110%), accuracy (99.34%) and F-Measure (97.90%). The results illuminate the better performance than existing anomaly detection techniques in the big data environment. 
Keywords: Big Data, Deep learning, Hadoop, Intrusion Detection.
Scope of the Article: Deep learning