A difficult problem in the realm of intrusion detection systems (IDS) is estimating the progress made in the identification of malicious code. Machine learning IDS training is dependent on the datasets provided, but gathering a valid dataset for comparison is difficult. To begin with, it is difficult to compare datasets since there is no standard approach for doing so, and also because there aren't any ground-truth labels or publicly available or real-world environment traffic, among other things [2]. Furthermore, only a few statistics reflect the current state of network traffic, which is almost exclusively encrypted for the sake of communication security and privacy. In the proposed system, a dataset is employed that satisfies both the content and the process requirements. The hybrid system for intrusion detection using data approach was introduced in the suggested study. Cybercrime is committed by a malicious node that can be identified by these tools. The goal of this research is to identify the most relevant and useful attributes for inclusion in a new IDS dataset. An approach for producing optimal ensemble IDS is devised in order to meet the goal. Information Gain (IG), Gain Ratio (GR), Symmetrical Uncertainty SU, Relief-F (R-F), One-R (OR) and Chi Squared are utilised and compared (CS). Techniques that use feature selection produce a list of the features that have been prioritised. For each of the four classification methods, we trained three other models on three different datasets for scanning and DDoS attacks and compared their performance with the proposed approach. In comparison to other trained models, the results of the experiments show that the proposed approach is more effective in preventing and detecting botnet attacks.

Intrusion Detection System(IDS)