MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202611106907 A) filed by Panipat Institute Of Engineering And Technology Piet on September 04, 2026, for A System For Proactive Software Fault Prediction Using Synthesized Domain Indices And Explainable Machine Learning.
Inventors include Dr. Gurmeet Kaur; Prof. Dr. Dinesh C. Verma; and Dr. Noopur Tyagi.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: 050] A processor-implemented system for proactive software fault prediction is disclosed. Software product and process metrics obtained from one or more repositories are cleaned, scaled and statistically filtered using correlation analysis and analysis-of-variance based feature selection. Retained metrics are synthesized into a complexity-stability ratio representing structural vulnerability and an operational maturity score representing lifecycle execution capacity, and the indicators are normalized to a bounded range. A regularized tree-based ensemble processes the indicators to track non-linear fault-risk boundaries and can update normalized sample weights to emphasize volatile or poorly predicted software blocks. A validation and explainability interface evaluates model performance using error metrics and multi- fold cross-validation and determines game-theoretic local contribution values. An output module supplies a fault-risk profile and diagnostic information to a software- engineering control environment for pre-deployment review or targeted refactoring. Accompanied Drawing [FIGS. 1-2]
Disclaimer: Curated by HT Syndication.
