Open AccessPerformance Analysis of Proposed Hybrid Machine Learning Model for Efficient Intrusion Detection
Author(s) -
Aditya Harbola*,
Priti Dimri,
Deepti Negi
Publication year - 2020
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.e3467.049620
Subject(s) - computer science , intrusion detection system , backdoor , feature selection , the internet , machine learning , artificial intelligence , data mining , network security , attack model , feature (linguistics) , anomaly based intrusion detection system , intrusion , computer security , world wide web , linguistics , philosophy , geochemistry , geology
At present networking technologies has provided a better medium for people to communicate and exchange information on the internet. This is the reason in the last ten years the number of internet users has increased exponentially. The high-end use of network technology and the internet has also presented many security problems. Many intrusion detection techniques are proposed in combination with KDD99, NSL-KDD datasets. But there are some limitations of available datasets. Intrusion detection using machine learning algorithms makes the detection system more accurate and fast. So in this paper, a new hybrid approach of machine learning combining feature selection and classification algorithms is presented. The model is examined with the UNSW NB15 intrusion dataset. The proposed model has achieved better accuracy rate and attack detection also improved while the false attack rate is reduced. The model is also successful to accurately classify rare cyber attacks like worms, backdoor, and shellcode.