Travis T Johnston
Research Scientist in Artificial Intelligence for High Performance Computing
Travis Johnston graduated with a Ph. D. in Mathematics in 2014 from the University of South Carolina where he studied problems in extremal combinatorics and graph theory. His work in machine learning and high performance computing began as a Postdoctoral Researcher at the University of Delaware while working in Michela Taufer's Global Computing Laboratory. He joined Oak Ridge National Laboratory as a Postdoctoral Researcher in machine learning and data analytics in July 2016. He has been a Research Scientist since June 2018. His research revolves around using high performance computing (HPC) to solve some of the most challenging problems related to artificial intelligence (AI) or Deep Learning (DL), machine learning, and data analytics. He is especially interested in model selection (e.g. how to pair the right neural network model with a dataset, also called hyper-parameter optimization), model selection and training with limited labelled data, and explainability (e.g. the ability ask a black-box neural network model why it made a particular decision). His research is powered by the fastest supercomputer in the world, Summit. His work combining AI and HPC earned him a coveted honor as a Gordon Bell Finalist at SC18.