MRI: Acquisition of a high-performance computing instrument to support deep learning, modeling/simulation, and visualization for STEM

University Of Maine, Orono ME

Investigators

Abstract

This project, acquiring an instrument for high performance computing, aims to support computational needs for deep learning, modeling, and visualization projects. The instrument is expected to drastically speed up projects that are now limited by available computational resources, as well as enable projects to be undertaken that currently cannot be done with existing resources in a reasonable amount of time, if at all. Deep learning training times can expect several orders of magnitude improvement, as can simulation and modeling, data mining, and preparation of visualizations and demonstrations. Deep learning models and scientific simulations that are too large to be run on current campus computing resources will be possible with the instrument. The following contributions are expected in this state, which is part of the Established Program to Stimulate Competitive Research (EPSCoR): - Scientific advances - for example, in understanding the influence of large-scale circulation patterns on physical structures and ecological habitats; the increased use of deep learning in STEM projects; - Emergence of new forms of understanding of data and models via emerging visualization techniques; - Outreach to K-16 education and the public through web-based dynamic visualizations; - Training of undergraduate and graduate students across science, technology, engineering, and mathematics (STEM) disciplines using the methodologies mentioned as well as the use of advanced computational tools; and - Enhanced access to computational resources in underserved rural communities throughout Maine. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.