The National Science Foundation awarded a $1,247,506 Project Grant to the University of Texas at Dallas under the Computer and Information Science and Engineering federal grant program (CFDA 47.070) for the period of October 1, 2022 through September 30, 2025.
The grant funds research to comprehend and mitigate errors in analog implementations of on-die neural networks. Specifically, the university will investigate and develop methods to address the impact of manufacturing and operational variations on machine learning models implemented through analog neural networks. Researchers will also develop techniques to evaluate the learning capacity of such designs and demonstrate the efficiency of proposed error mitigation solutions through custom analog neural network platforms. The work focuses on enabling robust and resilient operation of analog neural networks and the applications in which they are deployed.