Cornell University: Students win Qualcomm Fellowship with machine learning project
Doctoral students Chenhui Deng and Andrew Butt from Cornell’s School of Electrical and Computer Engineering have been awarded a 2022 Qualcomm Innovation Fellowship (QIF) for their proposal titled “Power Inference with Self-Supervised Learning.” Theirs was one of only 19 proposals selected from a highly competitive group of over 100 submissions.
“The goal of our work is to use machine learning techniques to enable fast and accurate power inference,” said Deng, the project leader. “Accurate power inference is critical to digital VLSI design, especially as existing computing systems from data centers to mobile devices become increasingly power constrained.”
Power inference in existing electronic design automation (EDA) tools is either inaccurate or very time-consuming. The team believes their proposed work will have a big impact in the EDA community, as it can not only tackle the challenges of power inference but also provide a novel and promising direction for researchers to solve other EDA problems.
Zhiru Zhang, associate professor of electrical and computer engineering, is advising the students.
“I’m excited about the prospect of this QIF project as it attempts to address one of the long-standing challenges of design automation of digital circuits by leveraging new advances in the field of machine learning,” Zhang said. “Chenhui and Andrew make a great team for tackling this problem, as Chenhui has strong expertise in machine learning on graphs and Andrew is well-versed in digital hardware design.”
The Qualcomm Innovation Fellowship program is focused on recognizing and rewarding innovative Ph.D. students across a broad range of technical research areas. QIF also enables graduate students to be mentored by company engineers and supports them in their quest towards achieving their research goals.
“This fellowship will provide us with invaluable expertise from Qualcomm engineers, especially in relation to which parts of circuits are most important for fast power inference,” Deng said. “Andrew and I hope this experience will also help us gain a deeper knowledge of which research problems are important in industry to guide us to impactful research in the future.”
The QIF program, open to students from a preselected variety of top U.S. based and international schools, receives over 100 proposals each year and has awarded over $15 million dollars since it started in 2009 at Qualcomm’s Research Center in Silicon Valley, California. Qualcomm Research’s top engineers carefully review submitted proposals and select the finalists, who are invited to present their proposals to a panel of executive judges. Winning students earn a one-year fellowship and are mentored by Qualcomm engineers to facilitate the success of the proposed research.