MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202611076706 A) filed by Noida Institute Of Engineering & Technology on June 20, 2026, for An Adaptive Hardware Acceleration Device For Machine Learning Tasks.
Inventors include Keshav Kumar Sharma; and Ayushi Gupta.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: An adaptive hardware acceleration device for machine learning tasks is disclosed. The device characterizes incoming workloads, evaluates policy constraints and operating telemetry, and configures specialized computation resources for suitable execution of dense, sparse, quantized, and data movement intensive operations. A scheduler cooperates with a profiling controller, memory traffic coordination logic, telemetry monitoring, and secure execution management to assign, adjust, and protect accelerated task segments during runtime. The device improves resource utilization, reduces transfer related delay, supports shared deployment, and adapts execution under changing workload conditions.
Disclaimer: Curated by HT Syndication.
