This study presents an Intrusion Detection System (IDS) that leverages machine learning and deep learning algorithms to enhance the detection and prevention of cyber threats in self-organizing networks. The proposed system integrates traditional machine learning techniques, such as Random Forest and Decision Tree, with advanced deep learning models, including Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, to improve detection accuracy across diverse network traffic data. To further enhance preprocessing efficiency, K-Best feature selection and K-Means clustering are utilized for dimensionality reduction and effective data categorization. A notable novelty of this approach is the incorporation of deep learning models, which are adept at handling large datasets with spatial and temporal patterns, and have demonstrated superior performance in detecting complex attack patterns. Additionally, a chatbot interface is integrated into the system to provide real-time intrusion detection information, allowing users to query and gain insights into detected threats. The backend is implemented using Python, while the frontend is developed with HTML, CSS, and JavaScript to ensure an interactive user experience. The experimental results show that while deep learning models such as CNN and LSTM offer high accuracy in detecting sophisticated attacks, traditional methods like Random Forest and Decision Tree excel in computational efficiency, making them ideal for real-time applications. The combination of these approaches ensures scalability, robustness, and adaptability in dynamic network environments.

Keywords: Intrusion Detection, Random Forest, Decision Tree, CNN, LSTM, K-Best, K-Means, Machine Learning, Deep Learning, Chatbot, Network Security, Flask.