MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202621084096 A) filed by Dr. Namrata Gain; Dr. Rachna Thakkar; Laxmi Madhu Kumar Brahmandam; Jeevan Krishna Paruchuri; Dr. Shubhangi Y. Chaware; Dr. Srinivasan Nagaraj; Mvvs Prasad; Dr. Ravi Kumar Vishwakarma; Dr. Ch. Raja; and Nagulapally Mamatha on July 08, 2026, for An Adaptive Decision Tree-Based Predictive Analytics System With Dynamic Feature Selection For Real- Time Data Processing.
Inventors include Dr. Namrata Gain; Dr. Rachna Thakkar; Laxmi Madhu Kumar Brahmandam; Jeevan Krishna Paruchuri; Dr. Shubhangi Y. Chaware; Dr. Srinivasan Nagaraj; Mvvs Prasad; Dr. Ravi Kumar Vishwakarma; Dr. Ch. Raja; and Nagulapally Mamatha.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: [060] The present invention relates to an adaptive hardware-implemented predictive analytics system for real-time data processing. The system comprises a stream acquisition interface, a feature conditioning module, a feature health monitoring module, a dynamic feature selection module, an adaptive decision tree prediction engine, a drift detection module, a tree update controller, a latency governance module, a memory management module, and an output interface. The feature health monitoring module evaluates live feature parameters including availability, delay, noise, stability, redundancy, computational cost, and prediction contribution. The dynamic feature selection module classifies features into active, reserve, and quarantine groups. The adaptive decision tree prediction engine generates predictions using active features and selectively updates affected decision paths without rebuilding the complete tree. The latency governance module controls feature count and tree traversal depth according to real-time processing budget. The system reduces latency, processor load, memory usage, and prediction failure during feature drift or feature unavailability.
Disclaimer: Curated by HT Syndication.
