Johanna Fröhlich represented the IDC at the NHR Conference 2026 in Paderborn, Germany. She presented a poster on her research, titled “The Confusion is Real: GRAPHIC – A Network Science Approach to Confusion Matrices in Deep Learning”. The work presents a new explainable AI (XAI) method for analyzing how neural networks distinguish between classes.
The NHR Conference brings together high-performance computing (HPC) users and providers from across the German National High-Performance Computing (NHR) centers. This year’s scientific program focused on HPC-driven decision science, atomistic simulations, and low- and mixed-precision computing for simulation and AI.