Information theft, forgeries, and denial of service are just a few of the threats and dangers that hackers and malware can pose to personal and business computer systems today. These actions can harm people and businesses greatly financially and morally. As a result, significant security precautions must be taken in this area. Many security measures have been suggested in various studies to guard against security flaws and threats. In this sense, the research initially examines the various types of network attacks before describing several defense strategies and intrusion detection systems. In the next step, the present research introduces an approach based on classification and feature selection, which is used to classify network traffic flows, i.e., normal traffic and attack traffic. In this regard, the input data is pre-processed and the SLPSO algorithm is used for feature selection. After feature selection, the decision tree algorithm is used to obtain a classification model to predict future attacks. Simulation result show that the proposed machine learning-based algorithm is able to represent the sequence of connections between computers in a network and can identify the types of network traffic well. In this method, by applying feature selection based on the SLPSO algorithm, network attacks are detected more accurately, which ultimately reduces the amount of classification error and increases the final accuracy compared to the previous article. So that the evaluation criteria such Precision, Recall, Accuracy and F-Measure between 17% to 20% improved compare the based method.
