MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202641082692 A) filed by Dr. C. Pratheeba; Mrs. R. D. S. Terese Sheeba; Mrs. N. S. Benila; Mrs. T. Thanka Geetha; Ms. A. Abira; Mr. C. Cibin; and Dr. J. R. Anisha on July 05, 2026, for A Method For Multi-Modal Phishing Email Detection Through Stacked Ensemble Learning Integrating Semantic Language Understanding, Structural Metadata Analysis And Unsupervised Anomaly Detection.
Inventors include Dr. C. Pratheeba; Mrs. R. D. S. Terese Sheeba; Mrs. N. S. Benila; Mrs. T. Thanka Geetha; Ms. A. Abira; Mr. C. Cibin; and Dr. J. R. Anisha.
The application for the patent was published on July 10, 2026, under issue no. 28/2026.
Abstract: The present invention relates to a multi-modal phishing email detection framework integrating semantic language understanding, structural metadata analysis, and unsupervised anomaly detection through stacked ensemble learning. The framework utilizes a DistilBERT transformer model for semantic phishing analysis, a Gradient Boosting Machine classifier for metadata-driven structural classification, and an Autoencoder neural network for anomaly detection based on legitimate communication behaviour. Outputs generated by the analytical branches are combined using a logistic regression meta-learner within an out-of-fold stacked ensemble architecture. The framework generates phishing probability scores mapped into a four-tier operational risk classification mechanism including low risk, uncertain risk, high risk, and critical risk categories. The invention improves phishing detection accuracy, zero-day attack identification capability, semantic contextual understanding, anomaly detection robustness, cybersecurity interpretability, and enterprise deployment suitability while reducing false positive and false negative classifications in enterprise communication environments.
Disclaimer: Curated by HT Syndication.
