MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202641050728 A) filed by Ms. Sabitha K; Mr. Aashik Harishwar M L; Mr. Elavenil B; Mr. Sam Britto M; and Mr. Prasannaraj R on April 21, 2026, for Human Assisted Machine Learning For Cyber Threat Detection.
Inventors include Ms. Sabitha K; Mr. Aashik Harishwar M L; Mr. Elavenil B; Mr. Sam Britto M; and Mr. Prasannaraj R.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: ABSTRACT OF THE INVENTION: The growmg sophistication of cyber threats has made it necessary to rely on automated systems for effective detection and prevention. However, purely automated solutions often fall short when dealing with unclear or unfamiliar attacks that require ~- deeper contextual insight. The Human- Assisted Machine Learning for Cyber Threat 1--- · Detection approach combines-machine-learning techniques with human expertise~to build a hybrid security system. It continuously analyzes network logs to detect anomalies such as brute-force attempts, phishing activities, and port scanning. Identified threats are classified automatically-well-known attacks are mitigated using predefined responses, while uncertain cases are escalated to a web-based platform for human evaluation. The feedback provided by experts is stored and used to retrain the model, enhancing its performance and adaptability over time. This integrated method effectively combines automation with human judgment to deliver faster, more accurate, and flexible cybersecurity protection. -Q) C) Ill D.. Q) .~.... N The proposed system presents an advanced cybersecurity solution that evolves through ongoing collaboration between artificial intelligence and human analysts. In contrast to traditional detection tools that depend on fixed algorithms, this model adopts a feedback-driven architecture where each human intervention helps retine the system's accuracy. A web-based interface allows security professionals to examine flagged incidents, add contextual understanding, and annotate threat information, which is then incorporated into the model's continuous learning process. This repeated feedback loop ensures the system can adapt to new and previously unknown attack patterns. By merging real-time automated detection with expert insights, the system creates a self-improving security framework capable of addressing complex cyber threats with enhanced precision, efficiency, and resilience.ABSTRACT OF THE INVENTION: The growmg sophistication of cyber threats has made it necessary to rely on automated systems for effective detection and prevention. However, purely automated solutions often fall short when dealing with unclear or unfamiliar attacks that require ~- deeper contextual insight. The Human-Assisted Machine Learning for Cyber Threat 1--- · Detection approach combines-machine-learning techniques with human expertise~to build a hybrid security system. It continuously analyzes network logs to detect anomalies such as brute-force attempts, phishing activities, and port scanning. Identified threats are classified automatically-well-known attacks are mitigated using predefined responses, while uncertain cases are escalated to a web-based platform for human evaluation. The feedback provided by experts is stored and used to retrain the model, enhancing its performance and adaptability over time. This integrated method effectively combines automation with human judgment to deliver faster, more accurate, and flexible cybersecurity protection. -Q) C) Ill D.. Q) .~.... N The proposed system presents an advanced cybersecurity solution that evolves through ongoing collaboration between artificial intelligence and human analysts. In contrast to traditional detection tools that depend on fixed algorithms, this model adopts a feedback-driven architecture where each human intervention helps retine the system's accuracy. A web-based interface allows security professionals to examine flagged incidents, add contextual understanding, and annotate threat information, which is then incorporated into the model's continuous learning process. This repeated feedback loop ensures the system can adapt to new and previously unknown attack patterns. By merging real-time automated detection with expert insights, the system creates a self-improving security framework capable of addressing complex cyber threats with enhanced precision, efficiency, and resilience.
Disclaimer: Curated by HT Syndication.
