XSEDE Neocortex Grant Accepted!

The Pittsburgh Supercomputing Center (PSC), a member of the Extreme Science and Engineering Discovery Environment (XSEDE) consortium, has recently expanded their supercomputing systems to include a game-changing piece of hardware for machine learning applications: the CS-1 "wafer-scale" engine.  This massive processor, created by Cerebras, is optimized for tensor operations, the main type of mathematical operation used in machine learning programs.  Our grant proposes to apply an advanced graph theory technique called "Graph Neural Networks" (GNN) that combines network science and machine learning.  This GNN project will support ongoing brain imaging and brain network research we are conducting on the PSC Bridges-2 system, and ultimately help answer questions about brain differences underlying health and mental illness.