MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085227 A) filed by Vellore Institute Of Technology on July 11, 2026, for An Explainable Graph Neural Network (xgnn) Architecture For Real -Time Multi -Relational Urban Safety.
Inventor includes Vani Rajasekar.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: [0028] An Explainable Graph Neural Network (XGNN) architecture for real-time multi-relational urban safety topology mapping and adaptive geo-spatial routing is disclosed. The system comprises a Real-Time Geospatial Graph Construction Engine for transforming urban data into a multi-relational spatial graph with feature-embedded edges. A Multi-Relational Graph Neural Network Inference Engine models spatial topology and calculates risk propagation. A Cognitive Algorithmic Explainability Layer extracts influential features and optimizes an attribution mask for safety classifications, providing human-verifiable telemetry. An Adaptive Topological Cost Routing Matrix integrates safety indices and explainability attributes, dynamically adjusting routing based on user-selected modes (Safest, Balanced, Fastest). A Synchronized Dual-Pass Execution Pipeline concurrently calculates classifications and attribution masks, ensuring navigation updates within two seconds. [FIG. 1]
Disclaimer: Curated by HT Syndication.
