MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641112663 A) filed by Aditya Institute Of Technology And Management; Kurman Sangeeta; More Sukesh; S. Srinivas Rao; P. Sai Tharun; T. Srikanya; Reventh Raj Dasari; and Korla Swaroopa on September 20, 2026, for Machine-Learning-Based Autonomous Inference Verification Using Semantically Constrained Counter Examples.
Inventors include Kurman Sangeeta; More Sukesh; S. Srinivas Rao; P. Sai Tharun; T. Srikanya; Reventh Raj Dasari; and Korla Swaroopa.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: A machine-learning-based autonomous inference verification system (100) verifies an inference generated by an artificial intelligence model. An input data acquisition module (102) receives an input data object comprising a plurality of input features. An initial inference engine (104) processes the input data object using a machine- learning model to generate an initial inference. An uncertainty determination module (106) determines uncertainty associated with the initial inference. A counterexample generation library (108) generates semantically constrained candidate counterexample data objects. An adaptive counterexample selection engine (110) selects a candidate counterexample based on expected discrimination and modification cost. A counterexample processing engine (112) generates a counterexample inference. A prediction transition analysis engine (114) identifies a prediction transition and a decision-sensitive feature. An inference verification controller (116) determines an inference stability measure and generates a verified inference output based on the prediction transition and the inference stability measure.
Disclaimer: Curated by HT Syndication.
