Improved Machine Learning-based System for Intrusion Detection
Jiarui Feng · Advances in computer science research · 2024
In addressing the growing cyber threats prevalent in the digital age, this research presents Machine Intrusion Detection System Based on Learning (MLIDS), an innovative system for detecting intrusions that harnesses the power of deep learning through the integration of a Multi-Layer Perceptron (MLP).This study aims to enhance the precision of cyber attack detection while minimizing the occurrence of false positives.The MLP model, meticulously crafted using PyTorch, incorporates multiple hidden layers that effectively capture the intricate patterns and non-linear relationships embedded within network traffic data.Significant advancements, such as the implementation of adaptive learning rate modifications and the application of L2 regularization techniques, have substantially bolstered the model's ability to generalize across various scenarios.The empirical outcomes of this research are compelling, with MLIDS achieving an impressive detection accuracy of 98.76%, surpassing traditional methods such as Naive Bayes and Single-Layer Perceptrons.These results not only highlight the efficacy of MLIDS but also underscore the transformative potential of deep learning in the realm of cybersecurity.