Unveiling Network Anomalies: A Machine Learning Approach to Intrusion Detection
Anshika Sharma, Himanshi Babbar, Amit Kumar Vats, R Archana Reddy · 2024
This study investigates the efficacy of different machine learning(ML) models for detecting intrusions using the LUFlow dataset, which comprises comprehensive network traffic data. An Intrusion Detection System (IDS) is an essential component of network security, intending to detect the goal of detecting unauthorised entries or abnormalities that may signal hostile actions in this environment. Four major ML models: Logistic Regression (LR), Random Forest (RF), AdaBoost, and XGBoost models have been assessed all providing unique benefits in terms of interpretability, robustness, and performance. The dataset undergoes extensive preprocessing to guarantee that the data is error-free, pertinent to the task at hand, and appropriate for modelling purposes. This entails managing missing values, standardising numerical characteristics, coding variables with categories, and dividing the data into sets for training, validation, and testing. Preprocessing processes are crucial for enhancing the quality and dependability of the dataset, hence improving the accuracy and resilience of the models. By conducting a methodical assessment, the objective is to ascertain the most efficient ML algorithm for identifying intrusions in network data. The research findings have important implications for improving network security by offering a strong tool for detecting and preventing harmful activity in network environments at an early stage.