MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611066575 A) filed by Noida Institute Of Engineering And Technology Niet on May 27, 2026, for Hierarchical Graph-Based Retrieval Augmented Generation System With Adaptive Reinforcement Learning Integration.
Inventors include Sovers Singh Bisht; and Anamika Chaudhary.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: The present invention discloses a hierarchical graph-based retrieval augmented generation system (100) comprising a hierarchical knowledge graph module (110), a tri-stage retrieval engine (120) implementing inter-community filtering (121), community-level refinement (122), and entity-level search (123), a dynamic weighting reward reinforcement learning module (130) with adaptive weight controller (131) and multi-objective reward calculator (132), and a beam search optimized dynamic re-ranking module (140). The system (100) further comprises a hybrid hardware-software acceleration engine (150) including graph processing accelerator (151) and vector computation optimizer (152), a secure encrypted communication layer (160), and a fault-tolerant redundancy controller (170). The adaptive weight controller (131) dynamically adjusts reward weights based on optimization progress, preventing policy collapse and ensuring balanced multi-objective optimization. The system (100) achieves latency reduction exceeding eighty percent and accuracy improvements exceeding twenty percent compared to conventional approaches through hierarchical filtering, dynamic reward weighting, and hardware-accelerated processing.
Disclaimer: Curated by HT Syndication.
