MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641093918 A) filed by Vardhaman College Of Engineering on August 03, 2026, for Self-Evolving Conversational Agents Using Lifelong Learning And Memory-Augmented Networks.

Inventors include Mr. Manzoor Mohammad; Mr. Anudeep Meda; Mr. Chittireddy Amarendar Reddy; Ms. Koti Tejasvi; Ms. Depalle Sandhya Rani; and Ms. T Ishverya.

The application for the patent was published on August 14, 2026, under issue no. 33/2026.

Abstract: Self-Evolving Conversational Agents Using Lifelong Learning and Memory-Augmented Networks is the proposed invention. The invention disclosed herein proposes a self-evolving conversational agent that utilises lifelong learning and memory-augmented networks to provide adaptive, personalised, and context-aware human- computer interactions. The system analyses multimodal user inputs (e.g., text, speech, and contextual data) to infer user intent, emotions, and conversation goals. An Adaptive Hierarchical Memory-Augmented Lifelong Learning Network is a model that organises knowledge in four levels of memory: episodic, semantic, contextual, and procedural memory layers that enable efficient storage and retrieval of long-term conversational experiences. A memory consolidation engine evaluates new information using criteria such as relevance, novelty, frequency of use, temporal consistency, and confidence, and selectively updates the long-term knowledge repository, avoiding catastrophic forgetting. The dialogue generation module leverages transformer-based language understanding and retrieved memories to generate coherent, personalised and explainable replies. Reinforcement learning constantly enhances the quality of the conversation based on user feedback, federated learning, encrypted local memory and privacy-preserving mechanisms ensure secure knowledge evolution without revealing sensitive user data. The invention improves the conversational continuity, adaptability, learning efficiency, response accuracy, and personalisation for virtual assistants, healthcare, education, enterprise support, and intelligent customer service applications.

Disclaimer: Curated by HT Syndication.