MUMBAI, India, April 17 -- Intellectual Property India has published a patent application (202641043790 A) filed by Sr University, Warangal, Telangana, on April 6, for 'gnn based transformer intelligence multimodal ai framework for early disaster management and adaptive response using bioacoustics.'
Inventor(s) include Dannuri Mounika; Johnson Kolluri; and Dr. B. Sunil Srinivas.
The application for the patent was published on April 17, under issue no. 16/2026.
According to the abstract released by the Intellectual Property India: "The proposed system is made up of huge linked components that work together to construct a single creation. The data collection module meets various data with satellite imagery, environmental sensors, and bioacoustics sensors that captures the noises of more species, along with birds, insects, mammals, and amphibious. In a preprocessing technique, the collected data is elegant using techniques for feature origin, normalization, and noise purifying. To capture periodicity and temporal characteristics in audio data, signals are transformed into Mel spectrogram illustration. This work proposes a hybrid deep learning approach that combines Graph Neural Networks (GNN) and Transformer models to improve prediction accuracy. The GNN component captures spatial relationships between different geographical regions by modeling them as nodes and edges, enabling the system to understand how disasters spread across connected areas. The Transformer component analyzes temporal patterns in sequential data such as weather conditions, rainfall, and sensor readings using self-attention mechanisms. Essential element of the invention is the composite of Graph Transformer knowledge module. This component incorporates tuning neural networks for feature removal, graph neural networks for pattern relationships among species and ecological records, and transformer networks for recording long range temporal dependencies. The fusion of these models allows simultaneous training of spatial and temporal patterns, thus significantly enhancing prediction accuracy. The invention additionally incorporates a bioacoustics intelligence component that detects ecological anomalies by analyzing multispecies environmental sound. These deviations, which frequently lead catastrophic disasters, incorporate a typical vocalization pattern, the lack of patrol species, and disruptions in ecological symmetry. A complementary layer of early warning specifically not used in traditional systems is submitted by this ecological intelligence. Keywords Graph Neural Networks (GNN), Transformer Models, Multimodal AI, Bioacoustics Analysis, Disaster Prediction, Early Warning Systems."
Disclaimer: Curated by HT Syndication.
