MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202541041847 A) filed by Rajeshgotur; and Vandana Bhat on April 30, 2025, for Ai-Driven Hybrid Neural Network Architecture For Context-Aware Cybersecurity Anomaly Detection.
Inventors include Rajeshgotur; Vandana Bhat; Prof. K. Satyanarayan Reddy; and Dr Praveen B M..
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: ABSTRACT: This patent application discloses a novel AI-powered cybersecurity system for detecting sophisticated threats through a unified architecture integrating convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformer-based attention mechanisms. The system processes heterogeneous security data—including endpoint telemetry, network traffic, and system logs—via a multimodal fusion pipeline that normalizes, aligns, and encodes raw inputs into temporally structured feature tensors. A layered threat detection engine employs CNNs to extract spatial patterns, RNNs to model sequential attack behaviours, and transformers to identify long-range contextual relationships, enabling simultaneous analysis of spatial, temporal, and behavioural threat indicators. The invention introduces a dynamic context-aware scoring mechanism that prioritizes anomalies using adaptive risk thresholds derived from user roles, historical baselines, and environmental factors, significantly reducing false positives. A closed- loop feedback architecture continuously retrains models using analyst-validated alerts and emerging threat intelligence, ensuring adaptability to zero-day attacks. Technical innovations include a noise-resistant preprocessing pipeline for enterprise-scale data, L2- regularized hybrid model optimization for adversarial robustness, and real-time behavioural anomaly detection without predefined rules. This architecture advances cybersecurity technology by combining multi-model AI analysis, automated threat prioritization, and self-improving detection capabilities, offering measurable improvements in detection accuracy, operational efficiency, and scalability for modern network environments.
Disclaimer: Curated by HT Syndication.
