Enhancing Network Security with Deep Learning-Based Intrusion Detection Systems

Abstract

Due to the vast quantity of services that are provided to users online and the immense amount of digital private information that has been transferred in recent years, the majority of individuals now rely heavily on the internet in their everyday lives. On the other hand, as internet usage increases, so does the assault surface for cyberattacks. The internet will be considerably more susceptible if no effective defense mechanism is put in place, which will increase the likelihood that data will be compromised or leaked. This highlights the significance of deep learning approaches for intrusion detection systems (IDS) and how crucial IDS are to network security. Sophisticated attacks are typically difficult to detect for traditional IDS methods, which results in more false positives and undetected threats. To overcome these limitations, this work proposed an empirical detection system of different deep learning models i.e., Long-Short Term Memory (LSTM),Multi-layer Perceptron(MLP), Linear Support Vector Machine(SVM), Quadratic Discriminant Analysis, in these models, the LSTM model does best with an accuracy of 96 %, precision of 92%, recall score at 93%, and F-1 smart value as well get up to level with most traditional methods. Compared with traditional machine learning algorithms, deep-learning models are advantageous for IDSs to model complex and sequential data sets with results which showsthe improved detection rates while reducing false alarms and concluded that deep learning-based IDS can offer more steady and dependable security solution in unpredictable network environments.