MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202641087897 A) filed by Vellore Institute Of Technology on July 17, 2026, for Behavioral Graph Intelligence Framework For Autonomous Cyber Attack Detection And Containment.
Inventor includes Dr. Priyadharsini M.
The application for the patent was published on July 24, 2026, under issue no. 30/2026.
Abstract: ABSTRACT BEHAVIORAL GRAPH INTELLIGENCE FRAMEWORK FOR AUTONOMOUS CYBER ATTACK DETECTION AND CONTAINMENT A system for detecting and containing cyber attacks comprises a streaming graph construction engine (102) that receives raw audit events, classifies entities into processes, files, sockets, pipes, registry keys, and network entities, assigns typed edges, and incorporates entities and edges into a temporal provenance graph using a sliding-window technique with incremental updates. A behavioral anomaly detection module (104) processes the graph through a hierarchical graph attention network with entity-type- specific transformation matrices, computes node embeddings and anomaly scores based on deviation from baseline behavioral profiles, and classifies anomalies by attack lifecycle stages. A lateral movement prediction module (106) encodes causal attack paths using a temporal graph neural network with causal edge masking and generates ranked candidate next hop nodes. A kill-chain escalation module (108) applies Bayesian belief updates generating kill-chain stage confidence levels. An autonomous containment controller (110) selects and executes containment actions using reinforcement learning. (Fig. 1)
Disclaimer: Curated by HT Syndication.
