Predictive Modelling for Network Threat Detection using Artificial Intelligence Techniques

• Published in: 2025 5th International Conference on Pervasive Computing and Social Networking (ICPCSN) pp. 1299 - 1305
• Main Authors: P, Jonathan Paul, R, Manoranjitham
• Format: Conference Proceeding
• Language: English
• Published: IEEE 14.05.2025
• Subjects: Artificial Intelligence (AI); Computational modeling; Cybersecurity; Intrusion Detection Systems (IDS); IP networks; Machine learning; Machine Learning (ML); Machine learning algorithms; Network Threat Detection; Organizations; Pervasive computing; Predictive Modelling; Predictive models; Protocols; Social networking (online); Threat assessment
• DOI: 10.1109/ICPCSN65854.2025.11035811

This work presents the development of an improved network threat detection system that uses machine learning algorithms as artificial intelligence processing techniques. The system receives numerous information sources containing time stamp data and source IP address information and destination IP address data with protocol classification data and traffic record data in addition tio firewall log data and IDS/IPS alerts data and multiple device record data and user data. System inputs undergo analysis to generate anomaly scores which also provide security alerts that detail attacks alongside type information and severity ratings and recommended risk reduction actions. This project deploys Artificial Intelligence approaches to boost security findings speed and accuracy thus enabling organizations to improve their cybersecurity sectors. Researchers focus on developing secure security measures in altering environments by seeking the best machine learning algorithm that delivers effective threat detection alongside quick security responses. Researchers can use the "Cybersecurity Attacks" dataset on Kaggle to establish their working information base for this project that explores forthcoming automated network defense systems.