Graphs are pervasive in various fields, including social networks, biological interactomes, transportation networks, and more. Understanding these graphs through structured representations is crucial for grasping their complex topologies and interdependencies, which are vital for numerous applications within these domains. In recent years, Graph Neural Networks (GNNs) have attracted considerable attention and shown promising results in many graph learning tasks. Despite this progress, learning long-range dependencies within graphs continues to be a significant challenge. In this talk, Dr. Ma will discuss their recent advancements in enhancing the expressive power of GNNs to capture long-range information in graphs effectively, addressing real-world challenges in diverse interdisciplinary applications, such as medical image, brain network analysis, social network mining, and program analysis. She will also introduce innovative work in using graph machine learning for efficient workload partitioning in system optimization. This approach not only improves system performance but also supports the scalable distributed training of large-scale GNNs.
Join us for an enlightening session!
Speaker(s): Guixiang Ma,
Virtual: https://events.vtools.ieee.org/m/417881
