Machine Learning (ML) and Deep Learning (DL) technologies are gaining prominence, particularly in computer network security. Cutting-edge technologies, such as fraud detection systems, network anomaly detection tools, and intrusion detection solutions, are now widely used for various applications. Over several decades, anomaly detection, also known as outlier identification, has been a persistent and active study topic in different academic areas. Specific problems with unique complexities and challenges still necessitate sophisticated solutions. Deep anomaly detection using deep learning has been a significant field of research in recent years. The combination of machine learning, deep learning, and cognitive intelligence gives a strong framework for enhancing network security and defending key assets from emerging cyber threats. The primary objective of this study was to showcase how the Ensemble model can enhance the effectiveness of a DL model, particularly an autoencoder, in detecting anomalies in computer networks. We increased the performance of the DL model by evaluating its findings. The experiment was conducted on a subset of the most recent CICIDS2017 network dataset. The Ensemble model attained an excellent overall accuracy of 99.74% and 0.99, respectively.

Deep Learning, Unknown Attack Detection, Autoencoder, Computer Network