Because of the constant dangers and assaults to which we are subjected, network security is essential to our everyday existence. As a result, many defense strategies and tactics must be developed. Systems for detecting network intrusions can be used to recognize an assortment of hostile network assault. In order to detect attack variations, a lot of systems now in use have concentrated on creating intrusion detection systems that make use of machine learning (ML) methods. By analyzing the characteristics of a sizable dataset, machine learning techniques may automatically identify the key distinctions among normal and aberrant data. It is true that a large number of features can be extracted without discriminating, which makes computing more complex. In order to enhance the efficiency of machine learning (ML) dependent detection techniques, a subset of features is chosen from the entire feature set using a feature selection method. A nature-based optimization technique that has shown effective in reducing processing obstacles to carry out feature selection problem optimization is the SALP swarm algorithm (SSA). The suggested method uses a Naïve Bayes classifier to examine how the SSA might enhance machine learning-based network anomaly identification.