MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202611076878 A) filed by Abes Engineering College on June 22, 2026, for Entropy-Regularized Self-Supervised Cyber Threat Detection And Adaptive Response Framework Using Deep Learning For Zero-Day Attack Mitigation.

Inventors include Mr. Vikas Maurya; Ms. Anshika Chaudhary; and Ms. Yashi Rastogi.

The application for the patent was published on August 07, 2026, under issue no. 32/2026.

Abstract: The present invention relates to an entropy-regularized self-supervised cyber threat detection and adaptive response framework for detecting, analyzing, prioritizing, and mitigating cybersecurity threats, including previously unseen zero-day attacks. The framework acquires multimodal cybersecurity telemetry from heterogeneous digital environments and processes the acquired data to generate standardized behavioral representations. A self-supervised learning mechanism learns normal cybersecurity behavior from unlabeled data, while an entropy-regularization mechanism improves representation robustness, uncertainty quantification, and anomaly discrimination. An entropy-based anomaly detection mechanism identifies suspicious cybersecurity activities and generates anomaly scores indicative of potential threats. Detected threats are classified and prioritized based on contextual cybersecurity information and threat severity indicators. A reinforcement-learning-based adaptive response mechanism autonomously determines optimized mitigation strategies and executes corresponding cybersecurity defense actions. The framework further incorporates a continuous feedback and model optimization mechanism configured to retrain and improve threat detection and response performance. The invention provides improved zero-day threat detection, reduced false-positive alerts, intelligent threat prioritization, autonomous mitigation, and continuous cybersecurity adaptation against evolving attack landscapes. FIG. 1, FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 6

Disclaimer: Curated by HT Syndication.