MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541122401 A) filed by S. Hrushikesava Raju; A. Anil Kumar; and Koneru Lakshmaiah Education Foundation, Mangalagiri, Andhra Pradesh, on Dec. 5, 2025, for 'a hybrid architecture integrating adaptive rules and dynamic behavioral learning for user profiling using multi-fusion.'
Inventor(s) include A. Anil Kumar; and S. Hrushikesava Raju.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "Most traditional methods suffer from poor interpretation of heterogeneous data, handle static behavior patterns, and are unable to capture dynamic patterns, resulting in a lack of real-time support, high false positives and false negatives, and a reliance on handcrafted rule-based systems, which are expected to evolve into adaptive rule systems for effective user profiling. Hence, a demanding scenario presents a hybrid risk-assessment framework for evaluating user profiles by integrating static attributes, dynamic behavioral patterns, and adaptive rule-based constraints. The model combines a deterministic Rule Engine for enforcing essential security and policy conditions, an LSTM/GRU network for learning temporal interactions and evolving user behavior, and an XGBoost classifier for capturing complex relationships within profile features. The outputs of these models were merged by a Fusion Layer, enabling them to generate the risk scores of a user. This multi-layered architecture ensures better accuracy, handles continuous learning from user interactions, and provides a robust, adaptive mechanism for real-time risk profiling. The hybrid approach strengthens decision-making, reduces false positives, and provides a scalable solution suitable for dynamic and data-driven risk assessment environments."
Disclaimer: Curated by HT Syndication.
