Skip to main content

Artificial Intelligence/Machine Learning

Cornell is a recognized leader in AI. Cornell researchers depend on CAC systems and consulting to enable AI and ML application innovations. See our AI/ML Services poster and AI/ML Services presentation to learn more. Below are a few examples of CAC-supported projects.

  • Matt Marx, Dyson, uses CAC systems to link patents to academic articles to understand the scientific heritage of innovation. Marx has used Red Cloud and plans to use a new CAC system that spawns 8 64-core AMD EPYC nodes in a virtual cluster. Marx has combined hand-tuned heuristics and the GROBID machine-learning package to achieve much higher performance than machine learning alone.
  • Amy Kuceyeski, Professor of Mathematics in Neuroscience at the Feil Family Brain & Mind Research Institute at WCM helped to organize an Intercampus Symposium on Machine Learning in Medicine and, most recently, ran a 128-core instance in Red Cloud for over two weeks to support her research in quantitative neuroimaging of neurological disorders.
  • Robert A. DiStasio Jr., Chemistry & Chemical Biology, runs simulations and machine learning on molecular properties and chemical reactions using the POOL Cluster built and maintained by CAC. A Slurm partition provides access to large-memory nodes; the largest has 1.5TB RAM and 7TB scratch.
  • Sara C. Pryor, Earth & Atmospheric Sciences, leveraged CAC's $8.2 million NSF-funded Aristotle cloud computing project. Pryor's use case and 6 others from Cornell, UB, and UCSB resulted in 147 publications during the life of the project. Pryor is now investigating whether ML can improve the forecasting of wind gust occurrence and magnitude.
  • Carla Gomes and Bart Selman, CS, lead the Institute for Computational Sustainability whose multidisciplinary researchers use the Institute's ATLAS2 HPC Cluster to develop constraint optimization, machine learning, and dynamical models for computational sustainability.
  • Ziv Goldfeld, Electrical & Computer Engineering, is providing first-of-a-kind performance guarantees for neural estimates of statistical distances that may lead to advances in machine learning. Goldfeld uses the Red Cloud platform.
  • Guy Hoffman, Mechanical & Aerospace Engineering, uses Red Cloud to train ML models for robot perception in the context of Human-Robot Interaction.
  • Murillo Campello, Johnson, is proposing a machine learning approach to Merger & Acquisition outcome prediction using Red Cloud computing and storage.
  • Parminder Basran, Veterinary Medicine Clinical Sciences, has a keen interest in ML methods in radiation oncology. CAC prepared scripts and demoed how MATLAB PCT works on a local machine and Red Cloud, and provided workflow integration advice.
  • Eun-Ah Kim, Physics, pioneered applying machine learning to quantum matter data. CAC built a Docker container and the Kim Group ran over 1 million hours on Red Cloud. CAC also previously maintained a cluster with GPUs.
Image designed by Starline