MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202621072836 A) filed by Symbiosis International Deemed University on June 11, 2026, for Explainable Ai Framework For Decision Failure Risk Estimation Using Stacked Ensemble Learning And Dual Interpretability.
Inventors include Gitika Makheja; Harman Saini; Saideep Pillay; and Dr. Gagandeep Kaur.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT EXPLAINABLE AI FRAMEWORK FOR DECISION FAILURE RISK ESTIMATION USING STACKED ENSEMBLE LEARNING AND DUAL INTERPRETABILITY The present invention relates an explainable artificial intelligence framework (100) for credit risk prediction and decision failure risk estimation. The framework integrates a stacked ensemble module (170) comprising Logistic Regression (161), Random Forest (162), XGBoost (163), and Multilayer Perceptron (164) as base learners with a meta-learner (171) trained on out-of-fold predictions. A data preprocessing module (120) with median imputation and a class imbalance handling module (130) using SMOTE prepare the financial data. The framework is characterized by a reliability scoring module (180) that computes prediction confidence using the formula Reliability equals the absolute value of P minus 0.5 multiplied by 2, and a dual explainability module (190) integrating SHAP (191) for global and local feature importance and LIME (192) for instance-specific explanations. The module (200) assesses demographic equity. The system achieves 87.96 percent accuracy with 0.8083 mean reliability and is deployed as an interactive application with automated PDF reporting (240). [
Disclaimer: Curated by HT Syndication.
