International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 05 | May 2025
p-ISSN: 2395-0072
www.irjet.net
Real-Time Cyber Threat Detection Using Deep Learning: A Step Towards Autonomous Security Ankush G Hegde 1, Anirudh P Nayak 2, Shrikara P S Nakshatri 3, Mrs. Geethapriya G H 4 1Student, Computer Science and Engineering, Jyothy Institute of Technology, Bengaluru, India 2Student, Computer Science and Engineering, Jyothy Institute of Technology, Bengaluru, India 3Student, Computer Science and Engineering, Jyothy Institute of Technology, Bengaluru, India
4Assistant Professor, Dept. of Computer Science and Engineering, Jyothy Institute of Technology, Bengaluru, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In today's rapidly evolving digital ecosystem,
Moreover, by reducing dependence on manual rule updates and security analysts, such systems can significantly lower response times and reduce operational overhead.
traditional rule-based cybersecurity systems are often ineffective against zero-day exploits and advanced persistent threats. This research presents an AI/ML-driven approach to intelligent intrusion detection and automated incident response. The proposed system captures real-time network traffic data, analyzes it using ensemble learning model trained on diverse attack patterns, and classifies potential threats with high accuracy. Upon detection, it generates context-aware mitigation recommendations and displays them via a userfriendly React-based dashboard. The solution helps businesses react to cyber threats more rapidly and precisely by utilizing data-driven automation and minimizing the need for manual monitoring. It improves the entire cybersecurity posture and adjusts to changing attack patterns through real-time detection and intelligent response. Results from experiments demonstrate its capacity to recognize various intrusion types and respond promptly, strengthening defenses against contemporary cyberthreats.
This research presents an automated cybersecurity incident response system that integrates a Machine Learning-based Intrusion Detection System (IDS) with realtime threat classification and response generation. By employing an ensemble learning model that has been taught using actual cases of normal traffic and various types of cyberattacks, the system monitors network traffic and calculates risks. With an intuitive dashboard designed using React, the system identifies the type of threat and presents possible responses upon detection of anomalous activity. Its flexible design provides transparent visual representations, the ability to learn and get better over time, and simple integration with existing security settings.
2. DEFINITION AND CONCEPTUAL FRAMEWORK
Key Words: Cybersecurity, Intrusion Detection System (IDS), Artificial Intelligence (AI), Machine Learning (ML), Real-time Threat Detection, Automated Incident Response, Network Traffic Analysis, Zero-day Exploits.
Through observing network traffic and recognizing abnormal or harmful activity, an intrusion detection system (IDS) has a significant role in cybersecurity. The two most popular forms of traditional IDS solutions are anomaly-based, which detect anything that is not normal behavior, and signature-based, which look for patterns that match documented threats. While beneficial, these methods often miss newer or unknown attacks, especially zero-day attacks, and need to be updated often to remain effective.
1.INTRODUCTION In the digital age, cyberthreats have increased in number and sophistication. People and companies are always at danger of serious data breaches or interruptions of vital services due to malware, phishing, ransomware, and unauthorized access assaults. Traditional security measures, which rely on pre-established guidelines and established threat patterns, are no longer sufficient to thwart complex, fast-moving, and advanced attacks, particularly those that are novel or unidentified, such as multi-layer intrusions and zero-day vulnerabilities.
This study employs ensemble learning model to track and classify network traffic in real time using Artificial Intelligence and Machine Learning to bypass the weaknesses of traditional systems. CNNs are extremely useful since they don't need human feature selection and can discover complex patterns in data automatically. With an easy-to-use dashboard built with React, the system offers real-time response suggestions when it detects suspicious activity. This approach minimizes the requirement for ongoing human supervision, streamlines threat response, and enhances detection accuracy by integrating automation and intelligence analysis.
To overcome these limitations, there is a growing need for adaptive, intelligent, and real-time security mechanisms. Artificial Intelligence (AI) and Machine Learning (ML) have shown great promise in enhancing cybersecurity by enabling systems to learn from patterns, identify anomalies, and make data-driven decisions. This dynamic capability makes them particularly suitable for evolving threat landscapes.
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 881