Anomaly Detection for Network Security

Today, network security is crucial due to the rapid development of and internet technologies, as well continuous growth in threats. Detecting anomalies one approaches that may be used safeguard a network's security. Recent research has focused extensively on techniques for identifying abnormalities. Using Autoencoder model together with data pre-processing such resampling feature selection, this describes novel approach It been shown suggested strategy applicable intrusion data. A comparison reconstruction error threshold value determines whether traffic normal or anomalous. CICIDS2017 dataset selected evaluate implementation proposed based real-world, large-scale, current The pre-processed achieved F1-Score 76% which outperformed baseline without selection stages. This project investigated effect performance Autoencoder. At end project, it demonstrated methodologies are towards imbalanced