A Hybrid Approach for Network Intrusion Detection

Due to the widespread use of internet and smart devices, various attacks like intrusion, zero-day, Malware, security breaches are a constant threat any organization's network infrastructure. Thus, Network Intrusion Detection System (NIDS) is required detect in traffic. This paper proposes new hybrid method for intrusion detection attack categorization. The proposed approach comprises three steps address high false low false-negative rates In first step, dataset preprocessed through data transformation technique min-max method. Secondly, random forest recursive feature elimination applied identify optimal features that positively impact model's performance. Next, we Support Vector Machine (SVM) types Adaptive Neuro-Fuzzy (ANFIS) categorize probe, U2R, R2U, DDOS attacks. validation calculated Fine Gaussian SVM (FGSVM), which 99.3% binary class. Mean Square Error (MSE) reported as 0.084964 training data, 0.0855203 testing, validate multiclass