MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641127079 A) filed by Dr. E. Sahaya Chithra; Ms. R. Kalpana; Mrs. Annabel Shimi S P; Dr. G. Indra Navaroj; Mrs. J. Gold Beulah Patturose; Ms Bebisha Therese G; Dr. N. Nisha Rosebel; and Mr. S. Mohamed Yas Nivisan on October 26, 2026, for Intelligent Blockchain Framework For Secure Deep Learning Model Protection And Integrity Validation.
Inventors include Dr. E. Sahaya Chithra; Ms. R. Kalpana; Mrs. Annabel Shimi S P; Dr. G. Indra Navaroj; Mrs. J. Gold Beulah Patturose; Ms Bebisha Therese G; Dr. N. Nisha Rosebel; and Mr. S. Mohamed Yas Nivisan.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: Traditional model protection methods tend to be based on the centralized security model, potentially leading to single points of failure, reduced traceability and challenges related to the authenticity of models in distributed environments. This study offers an approach to an Intelligent Blockchain Framework for Secure Deep Learning Model Protection and integrity validation, which combines blockchain technology, cryptographic verification and intelligent deep learning security mechanisms to create a tamper-resistant model management architecture. The proposed framework utilizes cryptographic hashing to create a unique digital fingerprint of trained models and also includes integrity metadata, version history, and model ownership details on a permissioned blockchain. Smart contracts automatically register models, grant access rights, and verify their integrity and detect suspicious modifications. In addition, an intelligent anomaly detection module scrutinizes model access patterns and updates and verification events to detect potential malicious actions. A secure model validation mechanism that verifies the computed model fingerprint against its registration on the blockchain prior to deployment and execution, preventing malicious modifications from being detected. The framework features off-chain model storage, encrypted model transfer, and role-based access control while preserving verifiable on-chain audit records, improving scalability and ensuring confidentiality. The purpose of the proposed architecture is to enable more authentic, traceable, tamper-detecting, and accountably manageable deep learning models through their lifecycle. The effectiveness of its performance can be assessed by integrity verification accuracy, the rate of detecting unauthorized modifications, transaction latency, computational overhead, storage efficiency, and access-control effectiveness. The framework provides a promising basis for deploying deep learning in a trusted manner across various application domains where security implies the need for protection of models and verifiable integrity, such as the healthcare sector and industrial automation.
Disclaimer: Curated by HT Syndication.
