MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085266 A) filed by M. Sivajyothi; and Dr. A. Raji Reddy on July 12, 2026, for Method And System For Verifiable Causal Learning Using Intervention- Based Data Augmentation And Attribution Constraints.
Inventors include Dr. P. Swetha; Dr. K. Srujan Raju; V. Sandya; K. Anoosha; B. Prashanth; and P. Rashmitha.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: A system (100) and method for verifiable causal learning are disclosed. The system includes a memory (102), a processor (104), and a communication module (106), and is configured to train a machine learning model using intervention-based data augmentation and attribution constraints. An intervention generation module (108) produces augmented data by modifying target variables based on a causal structure. An attribution computation module (110) determines feature attribution scores for both observational and augmented data. An attribution consistency constraint module (112) enforces constraints to align attribution behavior with causal relationships. A verification module (114) computes a causal consistency metric and generates a verification signal indicating causal validity. A model update module (116) iteratively updates model parameters based on augmented data, enforced constraints, and the verification signal, while a control module (118) dynamically adjusts training parameters. The disclosed framework enables closed-loop training and verifiable causal consistency.
Disclaimer: Curated by HT Syndication.
