| Title |
Class-focused evaluation of deep learning techniques for network intrusion detection / |
| Authors |
Bacevičius, Mantas |
| DOI |
10.15388/Proceedings.2024.44 |
| Full Text |
|
| Is Part of |
IVUS2024: 29th international conference "Information society and university studies", Vilnius University, Kaunas Faculty, Kaunas, Lithuania, May 17th, 2024: abstracts.. Vilnius : Vilniaus universiteto leidykla. 2024, p. 36 |
| Abstract [eng] |
In an increasingly interconnected world, safeguarding digital systems and networks against cyber threats is of utmost importance. Traditional intrusion detection approaches, relying on rule-based systems or simplistic machine learning models, often struggle to adapt to the evolving threat landscape. Deep Neural Networks (DNNs) offer promising avenues for enhancing Intrusion Detection Systems (IDS) effectiveness, leveraging their hierarchical structure to process complex network traffic data and extract discriminatory features indicative of malicious activity. However, the temporal dynamics inherent in network traffic data pose a unique challenge, prompting exploration into Long Short-Term Memory (LSTM) networks, for their sequential data processing capabilities. This paper investigates the application of deep learning models, including dense neural networks and LSTMs, for classifying network traffic into 28 distinct attack types. By analyzing the architectural design and presenting experimental results on standard benchmark datasets, we demonstrate the practical applicability of our hybrid approach in real-world cybersecurity scenarios, contributing to the advancement of intrusion detection systems through deep learning techniques. Additionally, we explore the challenges posed by class imbalances and dataset characteristics, providing insights into model performance and limitations for various attack types. |
| Published |
Vilnius : Vilniaus universiteto leidykla |
| Type |
Conference paper |
| Language |
English |
| Publication date |
2024 |
| CC license |
|
