Xiaofan Zhang has been selected as one of 30 awardees of a 2020 Google Ph.D. Fellowship across North America and Europe. The award recognizes outstanding graduate students who have done exceptional work in a number of computing disciplines. The Google Fellowship offers two years of support, a $35K annual stipend, full coverage of tuition and fees, and the opportunity to work with a Google research mentor.
"The student nominations we received were exemplary in their quality, but Xiaofan especially stood out and was endorsed by the research scientists and distinguished engineers
within Google who participated in the review," wrote the Google Ph.D. Fellowship committee. "We have given these students unique fellowships to acknowledge their contributions to their areas of specialty and provide funding for their education and research. We look forward to working closely with them as they continue to become leaders in their respective fields."
Zhang is advised by CSL Prof. Deming Chen in the ES-CAD research group. He has been working with CSL Prof. Wen-mei Hwu, Prof. Jinjun Xiong, and Dr. Cong (Callie) Hao in the IBM-ILLINOIS Center for Cognitive Computing Systems Research (C3SR). Both Chen and Hwu are faculty in the Department of Electrical and Computer Engineering. Xiong is an IBM researcher and an adjunct research faculty in ECE, and Hao is a post-doc researcher in the ES-CAD group.
Zhang's research interests include deep learning, energy-efficient computing, hardware-software co-design, and FPGA. In his fellowship proposal, "Efficient Hardware/Software Co-Design Methods to Unleash Full AI Capabilities for Mobile Devices," Zhang aims to provide accessible, private, efficient, and high-quality AI services by using mobile devices. The major challenge comes from the gap between demanding AI applications and scarce on-device resources and power budgets. Since there is no perfect solution to solve this problem, Zhang proposes a hardware/software co-design approach to enable an ideal match between AI algorithms and their hardware accelerators to unleash AI capabilities from resource-constrained mobile devices.
Zhang has published the initial results of his proposal in six top-tier conferences. Among them, DNNBuilder won the IEEE/ACM International Conference on Computer Aided Design (ICCAD) William J. McCalla Best Paper Award in 2018 by delivering the state-of-the-art DNN (deep neural network) accelerators, and SkyNet won double championships of the extremely competitive System Design Contest for low power object detection in the IEEE/ACM Design Automation Conference (DAC-SDC) in 2019. Zhang will continue working toward his research goal to address the challenges of implementing AI applications on mobile devices with support from Google Fellowship.
To receive the Google fellowship, students must go through three rounds of evaluation. Students must be nominated by their department, and each department may submit up to four nominations. The Graduate College coordinates the campus-level selection process to determine the final nominees from Illinois and sent to Google for the national competition. This year, 30 awardees were selected across North America and Europe among a few hundred eligible nominees. The acceptance ratio of the national competition is less than 10%.
Since its inception in 2009, 11 other Illinois Ph.D. students have been named Fellows in the highly selective program; Zhang is the second awardee from ECE Illinois.