(IJACSA) International Journal of Advanced Computer Science and Applications, Vol. 16, No. 4, 2025 577 | Page www.ijacsa.thesai.org Enhancing Cybersecurity Through Artificial Intelligence: A Novel Approach to Intrusion Detection Mohammed K. Alzaylaee Department of Computing-College of Engineering and Computing, Umm AL-Qura University, Saudi Arabia Abstract—Modern cyber threats have evolved to sophisticated levels, necessitating advanced intrusion detection systems (IDS) to protect critical network infrastructure. Traditional signature- based and rule-based IDS face challenges in identifying new and evolving attacks, leading organizations to adopt AI-driven detection solutions. This study introduces an AI-powered intrusion detection system that integrates machine learning (ML) and deep learning (DL) techniques—specifically Support Vector Machines (SVM), Random Forests, Autoencoders, and Convolutional Neural Networks (CNNs)—to enhance detection accuracy while reducing false positive alerts. Feature selection techniques such as SHAP-based analysis are employed to identify the most critical attributes in network traffic, improving model interpretability and efficiency. The system also incorporates reinforcement learning (RL) to enable adaptive intrusion response mechanisms, further enhancing its resilience against evolving threats. The proposed hybrid framework is evaluated using the SDN_Intrusion dataset, achieving an accuracy of 92.8%, a false positive rate of 5.4%, and an F1-score of 91.8%, outperforming conventional IDS solutions. Comparative analysis with prior studies demonstrates its superior capability in detecting both known and unknown threats, particularly zero-day attacks and anomalies. While the system significantly enhances security coverage, challenges in real-time implementation and computational overhead remain. This paper explores potential solutions, including federated learning and explainable AI techniques, to optimize IDS functionality and adaptive capabilities. Keywords—Intrusion detection; machine learning; deep learning; zero-day attacks; anomaly detection; feature selection; reinforcement learning; cybersecurity I. INTRODUCTION Digital infrastructure growth during the past decades has elevated cybersecurity to become a vital concern which spans across all sectors. An increasing number of entry points in the computing environment resulting from growing system connectivity and widespread cloud adoption and rapidly expanding IoT deployments has intensified risk exposure [6]. The world witnessed over 5.5 billion record exposures through global data breaches in 2022 and cybersecurity experts predict this cybercrime will cost the world $10.5 trillion by 2025 (Cybersecurity Ventures, 2023). The static rule and signature-based IDS mechanisms used in traditional intrusion detection systems encounter difficulties in tracking down contemporary security threats [7]. Standard IDS systems create numerous erroneous alarms at a rate ranging from 20% to 30% while missing complex and new types of cyber attacks (Moustafa & Slay, 2022). The percentage of zero-day intrusions currently amounts to 10–15% of total cyber attacks so they represent a substantial detection blind spot for present-day security solutions (Alazab et al., 2023). AI-based intrusion detection systems (IDS) represent an optimal answer for security needs because they implement machine learning (ML) and deep learning (DL) technologies to detect security threats more effectively. Recent studies have highlighted the superior performance of machine learning models like Support Vector Machines (SVM) and Random Forests compared to traditional approaches, particularly in intrusion detection contexts [1]. The systems implement data- driven learning algorithms that enable the detection of emerging attack patterns and peculiar network activities which standard IDS cannot identify [3]. Support Vector Machines (SVM) with Random Forests and Extreme Learning Machines demonstrate excellent abilities to categorize managed data structures but Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) process unfiltered network traffic for sophisticated attack sign detection [2]. AI-based IDSs have progressed but they continue to encounter three main drawbacks which include excessive false alarms and deployment difficulties and significant processing requirements. The main strength of signature-based IDS solutions lies in their ability to identify known threats yet they struggle with discovering new threats. Anomaly-based IDS detects new threats effectively yet their capability to produce many false alarms negatively impacts operational efficiency [8]. A better detection framework needs to emerge due to advancing cyberattacks because it should offer high accuracy detection with low false-positive rates at all times. The research develops a combined AI-based intrusion detection system which unites ML and DL approaches for evaluation through benchmark datasets including UNSW-NB15 and NSL-KDD. The proposed model delivers detection results with 92.8% accuracy and 5.4% false positive rate alongside 91.8% F1-score which outperforms traditional IDS systems. The system implements SHAP-based feature selection for better interpretability and reinforcement learning for adaptive response which improves the overall system robustness [9]. The research outcomes from this study create significant impacts for security applications in the real world and academic research domains. The proposed system uses AI component