MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641108982 A) filed by St. Peters Engineering College on September 10, 2026, for A Meta-Learning-Based Adaptive Artificial Intelligence System For Rapid Model Personalization Across Domains.
Inventors include Mr. Rathod Koveed, Assistant Professor In Department Of Cse, St Peters; Mrs Karuna Valisammagari, Assistant Professor In Department Of Cseaiml, St Peters Engineering College, Opposite Ts Forest Academy; Mr. A. D. Sivarama Kumar, Assistant Professor In Department Of Cse, St; Mr. Siddartha Kotha, Assistant Professor In Department Of Cse, St; Mr. P. Anil Kumar Reddy, Assistant Professor In Department Of Cse, St; and Mrs. M. Usha, Assistant Professor In Department Of Cse, St Peters Engineering College, Opposite Ts Forest Academy Kompally Road, Dullapally, Maisammaguda, Medchal, Hyderabad, Telangana.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention discloses a Meta-Learning-Based Adaptive Artificial Intelligence System for rapid model personalization across different domains. Conventional artificial intelligence models are generally trained for specific datasets or tasks and may require substantial retraining, labelled data, computational resources, and manual parameter tuning when applied to a new domain. The proposed system addresses these limitations by learning reusable adaptation knowledge from multiple source domains and applying the learned knowledge to a target domain. The system comprises a data input module, preprocessing module, domain characterization module, meta-learning module, base artificial intelligence model, adaptive personalization module, performance evaluation module, feedback module, and knowledge repository. The domain characterization module analyzes target-domain properties, including data distribution, feature characteristics, task requirements, and available training samples. Based on the identified characteristics, the meta-learning module provides suitable adaptation knowledge and the adaptive personalization module selectively adjusts model parameters, features, or learning configurations for the target domain. The personalized model is evaluated using predefined performance measures, and the feedback module uses the evaluation results to refine the adaptation process. The knowledge repository stores domain characteristics, adaptation strategies, model parameters, and performance information for reuse in subsequent personalization tasks. The system can support limited-data and few-shot adaptation, thereby reducing dependence on extensive domain-specific datasets. The invention provides an automated and flexible approach for rapid cross-domain AI model personalization while reducing adaptation time, computational requirements, and manual intervention. The proposed system can be applied in healthcare, finance, education, cybersecurity, manufacturing, transportation, and other domains requiring efficient and reliable adaptation of artificial intelligence models.
Disclaimer: Curated by HT Syndication.
