Universal Grid-Graph Learning: Zero-Shot Transfer without Retraining or Fine-tuning

#ieee #power #energy #machin-learning #deep-learning #graphconvolutionalnetworks #transferability
Share

This talk introduces how to address a fundamental challenge in applying
deep learning to power systems: developing neural network models that transfer
across significant system changes, including networks with entirely different
topologies and dimensionalities, without requiring training data from unseen
reconfigurations. Despite extensive research, most ML-based approaches remain
system-specific, limiting real-world deployment. This limitation stems from a dual
barrier. First, topology changes shift feature distributions and alter input dimensions
due to power flow physics. Second, reconfigurations redefine output semantics and
dimensionality, requiring models to handle configuration-specific outputs while
maintaining transferable feature extraction. To overcome this challenge, we introduce
a Universal Graph Convolutional Network (UGCN) that achieves transferability to
any reconfiguration or variation of existing power systems without any prior
knowledge of new grid topologies or retraining during implementation. Our approach
applies to both transmission and distribution networks and demonstrates
generalization capability to completely unseen system reconfigurations, such as
network restructuring and major grid expansions.  Experimental results across
different power system applications, including false data injection detection and state
forecasting, show that UGCN significantly outperforms state-of-the-art methods in
cross-system zero-shot transferability of new reconfigurations.


  Date and Time

  Location

  Hosts

  Registration


  • Add_To_Calendar_icon Add Event to Calendar

Loading virtual attendance info...

  • 323 Dr Martin Luther King Jr Blvd
  • Newark, New Jersey
  • United States 07102
  • Contact Event Hosts
  • Co-sponsored by Power Systems Engineering Center (PSEC) at New Jersey Institute of Technology
  • Starts 22 October 2025 04:00 AM UTC
  • Ends 06 November 2025 05:00 AM UTC
  • 0 in-person spaces left!
  • No Admission Charge

  Speakers

Dr. Tong Wu of University of Central Florida

Topic:

Universal Grid-Graph Learning: Zero-Shot Transfer without Retraining or Fine- tuning

This talk introduces how to address a fundamental challenge in applying
deep learning to power systems: developing neural network models that transfer
across significant system changes, including networks with entirely different
topologies and dimensionalities, without requiring training data from unseen
reconfigurations. Despite extensive research, most ML-based approaches remain
system-specific, limiting real-world deployment. This limitation stems from a dual
barrier. First, topology changes shift feature distributions and alter input dimensions
due to power flow physics. Second, reconfigurations redefine output semantics and
dimensionality, requiring models to handle configuration-specific outputs while
maintaining transferable feature extraction. To overcome this challenge, we introduce
a Universal Graph Convolutional Network (UGCN) that achieves transferability to
any reconfiguration or variation of existing power systems without any prior
knowledge of new grid topologies or retraining during implementation. Our approach
applies to both transmission and distribution networks and demonstrates
generalization capability to completely unseen system reconfigurations, such as
network restructuring and major grid expansions.  Experimental results across
different power system applications, including false data injection detection and state
forecasting, show that UGCN significantly outperforms state-of-the-art methods in
cross-system zero-shot transferability of new reconfigurations.

Biography:

Tong Wu received the Ph.D. degree from the Department of Information
Engineering, The Chinese University of Hong Kong, Hong Kong, in 2021. He was a
Postdoctoral Associate at Cornell Tech, Cornell University, NY, USA, from 2021 to
2024. He is currently a tenure-track Assistant Professor with the Department of
Electrical and Computer Engineering, University of Central Florida, Orlando, FL,
USA. His research focuses on universal graph learning theory and algorithms for
control and optimization in networked systems, particularly power grid systems.

Email: