Intrusion Detection System Using Machine Learning Techniques for Increasing Accuracy and Distributed & Parallel Approach for Increasing Efficiency

In the current era of wide use of internet by the users across the globe, the data generated and its security both are the important issues. Hence, in any network to prevent the malicious activities from the intrusions, the early and correct detection of intruders is important. For that we are using Intrusion Detection systems (IDS), developed with many effective algorithms which are helping us to get rid of these attacks. But, sometimes these systems do respond with the detection of intruders who are in real, not the intruders, but the normal users. This makes these users difficult to continue with the normal communication and can be prevented from accessing the important services over the World Wide Web. Similarly, not tolerable at all, when it's false negative i.e. intruder is allowed to enter into network traffic being a normal user, as it may cost to an organization to the extent of being demolished their future. Though the percentage of these false positives and false negatives is less, it is creating issues in many important services. For that we can have a machine learning techniques which can effectively detect intrusions in the system without creating obstacles of false positives and false negatives, also we are proposing it for the online streamed data. Also, this IDS implementation using machine learning techniques is proposed with Distributed & Parallel approach for simultaneous processing of network traffic data by these different machine learning techniques. This approach enables us to implement this IDS design for Online streamed data for faster detection i.e. increasing efficiency of IDS.