MUMBAI, India, Oct. 12 -- Intellectual Property India has published a patent application (202641119208 A) filed by St. Peters Engineering College on October 05, 2026, for A Self-Evolving Deep Learning Framework For Autonomous Model Optimization And Continuous Knowledge Adaptation.
Inventors include Ms. Arukonda Srujana, Assistant Professor In Department Of Cseaiml, St Peters Engineering College, Opposite Ts Forest Academy; Mr. Vangalapudi Apurv, Assistant Professor In Department Of Cse, St; Mr. Prajwal N Nayak, Assistant Professor In Department Of Cse, St; Mr. Katha. Chandrashekhar, Assistant Professor In Department Of Cse; Mrs. M. Usha, Assistant Professor In Department Of Cse, St Peters; and Mr. Ls Abinash Nayak, Assistant Professor In Department Of Cseaiml, Malla Reddy Mr Deemed To Be University, Opposite Ts Forest Academy Kompally Road, Dullapally, Maisammaguda, Medchal, Hyderabad, Telangana.
The application for the patent was published on October 09, 2026, under issue no. 41/2026.
Abstract: The present invention relates to a self-evolving deep learning framework for autonomous model optimization and continuous knowledge adaptation. The framework is configured to enable a deep learning model to continuously monitor its operational performance, identify changes in newly available data, autonomously determine suitable optimization actions, and adapt learned knowledge according to evolving data and operational conditions. The framework comprises a data acquisition module, preprocessing module, deep learning model, performance monitoring module, data-distribution monitoring module, autonomous optimization module, knowledge adaptation module, model update module, validation module, and deployment module. During operation, newly available data is processed and evaluated to identify changes in data characteristics and model performance. Based on detected conditions, the autonomous optimization module determines suitable optimization actions, while the knowledge adaptation module incorporates newly acquired information into the learned knowledge of the model. An updated model is generated and validated against predetermined performance conditions before deployment. A feedback mechanism enables continued monitoring and repeated optimization and adaptation cycles during operation. The disclosed framework reduces dependence on repeated manual model redevelopment and enables intelligent learning systems to adapt continuously to evolving data and operational conditions, thereby supporting autonomous model optimization, continuous knowledge adaptation, and sustained model performance.
Disclaimer: Curated by HT Syndication.
