MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641112087 A) filed by Dr. Pvs Siva Prasad; K. Naresh Babu; Dr. Swati Mirlekar; Mrs. Boddu Sravan; Mrs. Chitty Praneeth Reddy; Ms. Kadamanchi Lavanya; Mrs. Pinninti Sunitha; and P. Jessie on September 18, 2026, for System And Method For Privacy-Preserving Federated Machine Learning For Industrial Predictive Analytics.
Inventors include Dr. Pvs Siva Prasad; K. Naresh Babu; Dr. Swati Mirlekar; Mrs. Boddu Sravan; Mrs. Chitty Praneeth Reddy; Ms. Kadamanchi Lavanya; Mrs. Pinninti Sunitha; and P. Jessie.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: [045] A privacy-preserving federated machine-learning system and method for industrial predictive analytics are disclosed. The system retains raw industrial telemetry at a plurality of industrial edge nodes and trains local predictive models at the nodes. Each local model update is associated with reliability, freshness, novelty, drift and data-quality information, and an adaptive privacy gateway selects a privacy treatment according to disclosure risk, remaining privacy budget and expected contribution utility. Protected updates are securely aggregated and fused using reliability- and privacy-aware contribution weights with bounded node influence. A controller monitors distribution or model drift, accumulated privacy expenditure and participant state, selectively triggers retraining, and validates a candidate global predictive model against industrial acceptance constraints before deployment. The framework thereby enables distributed predictive maintenance, quality forecasting and process analytics while reducing transfer of sensitive operational data and improving robustness to heterogeneous, stale, noisy or privacy-distorted model updates. Accompanied Drawing [FIGS. 1-2].
Disclaimer: Curated by HT Syndication.
