Best Use of High Performance Data Analytics & Artificial Intelligence
Readers’ Choice Awards
Scientists from Argonne National Laboratory, the University of Chicago, National Center for Supercomputing Applications and the University of Illinois at Urbana-Champaign introduced a novel set of practical, concise, and measurable FAIR (Findable, Accessible, Interoperable, Reusable) principles for AI models. They showcased their approach with a domain-agnostic computational framework that brings together the Advanced Photon Source at Argonne, the Materials Data Facility, the Data and Learning Hub for Science, funcX, Globus, the ThetaGPU supercomputer, and the SambaNova DataScale system at the Argonne Leadership Computing Facility. They combined Nvidia A100 GPUs, Nvidia TensorRT, Docker, Apptainer (formerly Singularity), and the SambaNova DataScale system to demonstrate the use of AI surrogates to enable accelerated and FAIR AI-driven discovery for high-energy diffraction microscopy. The work presents a domain-agnostic computational framework to enable autonomous AI-driven discovery at scale and is showcased in the context of accelerated high-energy diffraction microscopy.
Editors’ Choice Awards
Researchers at Carnegie Mellon University and the University of North Carolina at Chapel Hill are using XSEDE-allocated Bridges-2 at the Pittsburgh Supercomputing Center and Frontera at the Texas Advanced Computing Center – built by HPE and Dell respectively – to develop machine-learning-driven robotic production of MRI contrast agents. The CMU team created an “artificial chemist” that mimicked the expertise of human chemists, which in turn directed a robotic lab instrument at UNC to synthesize improved contrast agents for medical MRI imaging without human supervision. The algorithm narrowed a potential 50,000 polymers to a short list that, in laboratory tests, performed as much as 50 percent better than current MRI contrast agents, offering a path toward improving the sensitivity and specificity of MRI images.