Intellectual Property India Publishes Patent Application for A Machine-Learning-Based Active Inference Verification System And Method Using Adaptive Physical Perturbations

MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641112513 A) filed by Aditya Institute Of Technology And Management; Srinivasa Rao Konni; N. Saikiran; Pradeep Kumar Mahapatro; Dasaradha Arangi; M. Devamma; B. Kiran Kumar; and Badana Vasudevarao on September 18, 2026, for A Machine-Learning-Based Active Inference Verification System And Method Using Adaptive Physical Perturbations.

Inventors include Srinivasa Rao Konni; N. Saikiran; Pradeep Kumar Mahapatro; Dasaradha Arangi; M. Devamma; B. Kiran Kumar; and Badana Vasudevarao.

The application for the patent was published on September 25, 2026, under issue no. 39/2026.

Abstract: A machine-learning-based active inference verification system (100) for verifying an inferred physical state of a physical system includes a multi-modal sensor network (102) configured to acquire physical-state data. An initial inference engine (104) processes the physical-state data using a machine-learning model to generate candidate physical states and corresponding confidence values. An uncertainty determination module (106) determines inference uncertainty and identifies candidate states requiring discrimination. An active probe selection engine (110) selects a physical perturbation from a physical probe library (108) based on an expected discrimination value and perturbation cost. A physical excitation module (112) applies the selected perturbation, and a transient response acquisition module (114) acquires a resulting physical response. A response verification engine (116) compares a measured response signature with predicted response signatures. An inference verification controller (118) updates the confidence values and generates a verified physical-state inference.

Disclaimer: Curated by HT Syndication.