Intellectual Property India Publishes Patent Application for 'Neuro Symbolic Intelligence Architecture For Real Time Reasoning And Learning' Filed by Dr. S. Sivakumar; B Vishnupriya; Dr. M. Renuka; V Meena; Sivasamy C; and Dr. Gunasundari B

MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202641003806 A) filed by Dr. S. Sivakumar; B Vishnupriya; Dr. M. Renuka; V Meena; Sivasamy C; and Dr. Gunasundari B, Perambalur, Tamil Nadu, on Jan. 14, for 'neuro symbolic intelligence architecture for real time reasoning and learning.'

Inventor(s) include Dr. S. Sivakumar; B Vishnupriya; Dr. M. Renuka; V Meena; Sivasamy C; and Dr. Gunasundari B.

The application for the patent was published on March 13, under issue no. 11/2026.

According to the abstract released by the Intellectual Property India: "Neuro-symbolic intelligence has emerged as a promising paradigm to bridge the gap between data-driven leaming and logical reasoning. This work proposes a Neuro-Symbolic Intelligence Architecture for Real-Time Reasoning and Leaming that tightly integrates deep neural networks with symbolic reasoning mechanisms to achieve interpretable, adaptive,. and efficient decision-making. The architecture employs Cortvolmional Neural Netwud1.s. (CNNs) and : Transfonner-based encoders for high-level feature extraction from unstructured sensory data, while symbolic knowledge is represented using first-order logic rules and ontologies. A Differentiable Logic Layer (DLL) based on Probabilistic Soft Logic (PSL) enables seamless interaction between neural outputs and symbolic constraints, allowing end-to-end training via backpropagation. Real-time reasoning is achieved using an optimized forward-chaining inference engine coupled with incremental rule evaluation to minimize latency. For continual leaming, the system incorporates reinforcement leaming with policy gradients1 to update symbolic rules and neural parameters dynamically based on environmental feedback. The proposed architecture improves reasoning accuracy, leaming efficiency, and explainability compared to purely neural models, making it suitable for real-time applications such as autonomous systems, intelligent healthcare, and adaptive decision-support platforms."

Disclaimer: Curated by HT Syndication.