Experimental Analysis of various Machine Learning approaches for Intrusion Detection

Intrusion Detection System (IDS) is basically a system that monitors the system or network activities and detects the occurrence of malicious activities. Intrusion detection is important as it plays a significant role in developing an efficient system. Anomalies need to be identified as they can lead to the root cause of an issue or mishap by analyzing the data and also reduce the threats and vulnerabilities to the system. Machine learning algorithms work better as they are adaptive and can give time analysis and insights into the test data. The algorithms can be trained effectively for multiple data and can track the exploitation of the system. Numerous approaches can be followed for intrusion detection. This paper discusses and compares four major supervised machine learning algorithms and how they are effective in various test conditions. The paper aims to create an Intrusion Detection system comparing various ML techniques to distinguish between the normal (good connections) and the intrusions (bad connections), to recognize the best fit technique on which the further developments can be carried.