Design and Development of Intrusion Detection System Architecture by Using Ensemble Deep Neural Networks and Explainable Artificial Intelligence in Healthcare

Abstract

With the growing adoption of the Internet of Things (IoT) technology in healthcare, newlinethe risk of DDoS attacks has become more significant. In this thesis, a hybrid model newlinethat combines deep learning models such as Bidirectional LSTM (Bi-LSTM) with the newlineXGBoost algorithm for stacking ensemble to detect and prevent DDoS attacks in IoT newlinehealthcare environments with explainable AI (XAI) techniques to provide insights newlineinto the decision-making process of the proposed model, which can help network newlineadministrators to understand the root causes of DDoS attacks and take appropriate newlinemeasures to prevent them. The proposed model leverages the strengths of each newlinealgorithm to achieve robust accuracy and efficiency in detecting and preventing DDoS newlineattacks. newlineThe proposed approach uses the CICIDS 2019 dataset, which is a widely used dataset newlinefor DDoS attack detection. The experimental results show that the proposed model newlineoutperforms state-of-the-art methods in terms of accuracy, efficiency, and newlineinterpretability of the proposed model using LIME to provide insights into the newlinedecision-making process. Finally, discuss the limitations and challenges of the newlineapproach and propose directions for future research. This research contributes to the newlinedevelopment of effective and efficient DDoS detection and prevention systems for newlineIoT healthcare environments and provides insights into the decision-making process newlineof machine learning models using XAI techniques. newlineThe results of this research can help to improve the security of IoT healthcare newlineenvironments against DDoS attacks, contributing to the overall safety and well-being newlineof patients. newline