MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641075082 A) filed by Vardhaman College Of Engineering on June 17, 2026, for Efficient Edge Ai Processing Using A Hardware-Optimized Neural Network Accelerator.
Inventors include M. Nagarjuna; Sushanta Debnath; S. Rajendar; C. Padmini; R. Phani Vidyadhar; C. Srivardhini; Chenagani Samatha; Ganesh Chintale; and Madogoni Harika.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: Efficient Edge AI Processing Using a Hardware-Optimized Neural Network Accelerator is the proposed invention. The present invention relates to a hardware optimised neural network accelerator, designed for efficient real-time artificial intelligence processing at the edge. The proposed system aims for high performance inference in resource constrained environments, overcoming the limitations of conventional cloud-dependent and general-purpose computing solutions. It provides a certain parallel architecture that improves the basic functions of neural networks with reduced power consumption and latency. The accelerator employs optimised dataflow techniques and an effective on-chip memory architecture to minimise data transfer and enhance computational efficiency. To be efficient, the system is augmented with model optimisation techniques such as quantisation and pruning that allow small neural networks to run with little to no accuracy loss. The system can achieve energy-efficient operation by adapting the performance to workload requirements, based on adaptive precision computing and dynamic power management algorithms.
Disclaimer: Curated by HT Syndication.
