MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641051983 A) filed by Jisha Jose; and Dr. J. E. Judith, Thiruvananthapuram, Kerala, on April 23, for 'hybrid deep learning framework for real-time anomaly detection in iot data streams.'
Inventor(s) include Jisha Jose; and Dr. J. E. Judith.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "Aspects of present disclosure relate to hybrid deep learning-based anomaly detection in IoT data streams. The system integrates a Modified Harris Hawks Optimization (MHHO) algorithm for efficient feature selection with hybrid machine learning and deep learning architectures. The MHHO algorithm significantly reduces high-dimensional IoT traffic data while preserving critical features, improving computational efficiency and detection accuracy. The framework incorporates a stacked ensemble model combining Random Forest, XGBoost, and Support Vector Machine classifiers, along with Logistic Regression meta-classifier for robust anomaly classification. Additionally, a novel hybrid deep learning architecture (MHHO-CVTF-Net) integrates 1D CNN, BiLSTM, Variational Autoencoder, and Temporal Fusion Transformer to capture spatial-temporal dependencies in network traffic. The system demonstrates superior performance across benchmark datasets (IoT-23, IoTID20, CICIDS-2018), achieving high accuracy, low false positive rates, and strong generalization. The invention provides a scalable, adaptive, and efficient solution for real-time anomaly detection in IoT systems deployed at edge, fog, and cloud environments."
Disclaimer: Curated by HT Syndication.
