MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541123082 A) filed by Cvr College Of Engineering, Hyderabad, Telangana, on Dec. 6, 2025, for 'self-calibrating adversarially robust model compression framework with cross-modal certified defense for edge deployment.'
Inventor(s) include K Deepthi Reddy.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "Self-calibration adversarial robust model compression self-calibration An edge-deployable adversarial robust framework that combines structured pruning, mixed-precision quantization, and knowledge distillation with certified defense features. Multi-objective neural architecture search produces the Pareto-optimum compressed models in terms of size, latency, and L2 robustness assurances. Using randomized smoothing and interval bound propagation, verified through cross-modal contrastive losses, a certified defense layer trains to synchronize security in vision, audio, and text. A runtime calibration controller implements model dynamic switching and maintains metrical measurements of devices used to scale back and forward between the different model settings. This holistic framework has shown demonstrably secure, efficient AI models in resource-constrained hardware, in confidence in their use in self-directed systems, medical devices, and industrial IoT systems."
Disclaimer: Curated by HT Syndication.
