MUMBAI, India, May 29 -- Intellectual Property India has published a patent application (202611051381 A) filed by Noida Institute Of Engineering And Technology, Greater Noida, Uttar Pradesh, on April 22, for 'a system for continual learning based detection of ai-generated images using hardware-accelerated neural architecture.'
Inventor(s) include Anam Rajput; and Shweta Singh.
The application for the patent was published on May 29, under issue no. 22/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to a hardware implemented system (100) and method for detecting artificial intelligence generated images using a three stage continual learning architecture. The system (100) comprises a vision transformer processing unit (101) with low rank adaptation modules (102) fine tuning multi layer perceptron layers, a continual learning engine (103) employing an augmentation chain module (104) of progressively increasing complexity, a Kronecker factored approximate curvature estimator (105) computing Hessian approximations, a linear interpolation module (106) exploiting linear mode connectivity between parameter minima, a non transitory memory (107), and a hardware accelerator (108). The system (100) solves the technical problem of detecting synthetic images from evolving generative models while mitigating catastrophic forgetting. The invention achieves an average accuracy of 92.20 percent across 27 generative models with reduced memory footprint and processing latency, applicable in digital forensics, content authentication, cybersecurity platforms, and social media verification infrastructure."
Disclaimer: Curated by HT Syndication.
