MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641100040 A) filed by Mr. Debapratim Gupta on August 19, 2026, for Adaptive Communication-Efficient Federated Learning For Privacy- Preserving Customer Churn Prediction.
Inventor includes Mr. Debapratim Gupta.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: Adaptive communication-efficient federated learning for privacy-preserving customer churn prediction ABSTRACT The present invention relates to a privacy-preserving federated learning system and method for customer churn prediction across distributed travel service platforms. The system enables multiple travel service platforms to collaboratively train a global churn prediction model without sharing raw customer data. Each participating platform locally preprocesses customer-related interaction data and generates model updates. An adaptive multi-level communication mechanism dynamically selects an appropriate communication level for transmitting model updates based on model convergence, update significance, network conditions, communication cost, and prediction performance. A privacy-preserving aggregation mechanism securely combines the received updates to generate an improved global model while reducing the risk of customer information disclosure. The trained global model is distributed to participating platforms for local churn prediction and continuous model improvement.
Disclaimer: Curated by HT Syndication.
