Zhuo Feng Receives $800,000 NSF Grant to Investigate High-Performance Spectral Algorithms
Feng’s research will potentially boost the performance of emerging graph learning techniques.
Principal Investigator Zhuo Feng, associate professor in the Department of Electrical and Computer Engineering (ECE) at Stevens Institute of Technology, was recently awarded a grant of $800,000 from the National Science Foundation for his project “SHF: Medium: Co-optimizing Spectral Algorithms and Systems for High-Performance Graph Learning.” This is part of a four-year, $1.2 million collaborative project with Cornell University. Hang Liu, ECE assistant professor, is the co-PI of this project.
Emerging graph learning techniques have shown promising results for various important applications such as vertex (data) classification, link prediction (recommendation systems), community detection, drug discovery, and electronic design automation (EDA). However, even the state-of-the-art graph learning methods cannot scale to large data sets due to their high algorithm complexity.
This proposed research investigates high-performance spectral algorithms and systems for both (un)directed graphs and hypergraphs based on the latest theoretical breakthroughs. Unlike prior theoretical studies on spectral graph theory that put less focus on practical algorithm implementations and applications, Feng will for the first time develop practically efficient spectral compression techniques to dramatically reduce the sizes of real-world, large-scale (hyper)graphs without impacting structural properties. The proposed research, Feng envisions, will lead to the development of highly scalable graph processing systems taking full advantage of the latest heterogeneous computing devices,
The success of the proposed research will significantly advance the state of the art in spectral graph theory, machine learning, data analytics, electronic design automation (EDA), as well as high-performance computing (HPC).
Feng said, “Exploiting new spectral graph compression methods will allow us to achieve more efficient hardware-algorithm co-design of graph learning systems.”
