MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641126265 A) filed by Karpagam Academy Of Higher Education; Karpagam Institute Of Technology; Dr R. Subha; and Manju A on October 23, 2026, for Insider Threat Detection Using Machine Learning Model For User Behaviour Analysis.
Inventors include Dr R. Subha; and Manju A.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: Insider threats represent one of the most critical and challenging security risks faced by modern organizations, as they originate from trusted users who possess legitimate access to sensitive systems and data. With the rapid adoption of cloud computing, remote work environments, and Bring Your Own Device (BYOD) policies, traditional perimeter-based security mechanisms are increasingly inadequate for identifying malicious or negligent activities carried out by authorized insiders. This work presents a machine learning–based framework for insider threat detection through comprehensive user behavior analysis, aimed at enhancing proactive cybersecurity defense mechanisms. The proposed system continuously collects and analyzes user activity data from multiple sources, including system logs, network traffic, application usage records, and access control events. These heterogeneous data streams are preprocessed, normalized, and transformed into meaningful behavioral features that capture temporal, spatial, and contextual aspects of user interactions. By modeling normal user behavior patterns, the system establishes dynamic behavioral baselines that adapt to evolving user roles, responsibilities, and working conditions. Advanced machine learning techniques are employed to identify deviations from established behavioral norms that may indicate potential insider threats. Anomaly detection algorithms are utilized to uncover unusual activities that differ significantly from historical behavior, while clustering methods group users with similar activity patterns to enable peer-based comparison and contextual anomaly identification. In addition, deep learning models are incorporated to analyze high-dimensional data and detect complex, non-linear relationships associated with sophisticated insider attack strategies that are difficult to capture using traditional methods. To enhance decision-making and reduce false positives, the system integrates automated risk scoring and alert prioritization mechanisms. Detected anomalies are evaluated based on severity, frequency, and contextual relevance, allowing security administrators to focus on high-risk incidents that require immediate attention. The proposed framework is designed to be scalable and flexible, supporting deployment in cloud-based, distributed, and heterogeneous enterprise environments while enabling near real-time threat detection. Experimental evaluations demonstrate that the proposed machine learning–based approach significantly improves the accuracy and timeliness of insider threat detection compared to conventional rule-based security systems. By providing adaptive learning, continuous monitoring, and intelligent analysis of user behavior, the system strengthens organizational security posture and enhances resilience against insider-driven attacks. Overall, this work highlights the effectiveness of data-driven user behavior analysis in mitigating insider threats and safeguarding critical digital assets in modern cybersecurity landscapes.
Disclaimer: Curated by HT Syndication.
