Abstract: Deep learning has become increasingly vital in data science, especially when handling large datasets. This paper focuses on analyzing intrusion detection attacks, which are critical for maintaining information security. The core technology lies in accurately identifying various network attacks. We explore the development of an intrusion detection system based on deep learning and propose a method using recurrent neural networks (RNN-IDS) for this purpose. Our project involves analyzing the KDD dataset, which comprises 44 features. Utilizing these features, we apply an RNN classification algorithm to train the data and assess accuracy. We also compare our results with those obtained from decision trees, support vector machines, and other machine learning techniques used by previous researchers on the benchmark dataset. This comparative analysis aims to highlight the effectiveness of RNNs in intrusion detection.

Keywords: Deep Learning, Prediction, DoS, R2L, U2R, Prob attack.

| DOI: 10.17148/IARJSET.2024.11813