MACHINE LEARNING-BASED INTRUSION DETECTION SYSTEM FOR LOCAL NETWORKS

Abstract

The useful solution refers to a Machine Learning-Based Intrusion Detection System for intranets, comprising: at least one server including: data storage; and a processor configured to perform a process including the following steps: i) receiving input data; ii) performing merging, normalization, and feature extraction from the input data; iii) performing data enhancement and enrichment; iv) performing labeling of the input data in Step iii); and training to generate labeled data to identify and detect intranet intrusions; v) identifying and detecting intranet intrusions through the trained labeled data in Step iv) and network data streams connected to the intranet through a combination of CNN (Convolutional Neural Network) module, LSTM (Long Short-Term Memory) module; and GBM (Gradient Boosting Machine) module via Dynamic Weighted Voting (DWV) mechanism.and vi) displaying the results of detecting internal network intrusions and alerting users upon request via the user-connected device; b) the connection port is paired and electronically communicates between multiple user-connected devices and the server; and c) multiple user-connected devices connect and communicate with the server via the internet and connection port to monitor and receive alert information related to internal network intrusions.