Research Projects in Data-Enabled Industrial Mathematics

ERAU Research Experience for Undergraduates

Summer 2022 REU

INDUSTRIAL PARTNERS

Nevada National Security Site PNNL
Nevada National Security Site, Las Vegas, NV Pacific Northwest National Laboratory, Richland, WA

Projects

Quantifying Uncertainty in Ensemble Deep Learning
Industrial Sponsor: Nevada National Security Site, Las Vegas, NV
Description and results

Students:
Max Prilutsky (‘22) San Diego State University - Applied Mathematics
Emily Diegel ('24) DB BS Computational Mathematic
Rhiannon Hicks ('24) DB BS Astronomy & Astrophysics
Rachel Swan ('25) DB BS Computational Mathematic

Description: The Computing & Data Science Group at the Nevada National Security Site (NNSS) is pleased to partner with Research Experience for Undergraduates (REU) students at Embry-Riddle Aeronautical University (ERAU) to develop methods for quantifying uncertainty in neural networks to help improve network interpretability and correctness, using publicly available datasets.

Neural networks are an emerging topic in the data science industry due to their high versatility and efficiency with large data sets. The purpose of this modern machine learning technique is to recognize relationships and patterns in vast amounts of data that would not be explored otherwise. Past research has utilized machine learning on experimental data in the material sciences and chemistry field to predict properties of metal oxides. Neural networks can determine underlying optical properties in complex images of metal oxides and capture essential features which are unrecognizable by observation. However, neural networks are often referred to as a “black box algorithm” due to the underlying process during the training of the model. The explanation for a prediction is unable to be traced, therefore poses a concern on how robust and reliable the prediction model actually is. Building ensemble neural networks allows for the analysis of the error bars of the prediction model. The project’s objective is to determine the comparative differences between the predictive ability of each individual neural network versus the ensemble neural network. Additionally, the paper explores how to use the ensemble model as a method of uncertainty quantification. Overall, ensemble neural networks outperform singular networks and demonstrate areas of uncertainty and robustness in the model.

Major outcomes:

  • Results were presented at 2022 ERAU Discovery day
  • Results were presented at 2022 ERAU REU Showcase
  • Results were presented at 2022 SIAM Annual Meeting July 11-15, 2022
  • Results were presented at 2022 ERAU STUDENT RESEARCH SYMPOSIUM
  • Results were presented at 2023 Emerging Researchers National (ERN) Conference in STEM
  • Results were presented at JMM2023
  • Publication submitted to SIAM SIURO

Non-negative Matrix Factorization Approach to Computing “Fingerprints” in Spectra of Nuclear Materials
Industrial Sponsor: Pacific Northwest National Laboratory, Richland, WA
Description and results

Students:
ERAU REU: Research Projects in Data-Enabled Industrial Mathematics (Summer 2022)
Zachariah Kline ('23) Wisconsin Lutheran College - Mathematics and Computer Science
Emily Armstrong ('23) Assumption University - Mathematics
Jensen Bridges ('24) Oklahoma State University - Mathematics and Statistics
Zoe Friedman ('23) Illinois State University - Mathematics
Jacob Antici ('24) Arkansas State University - Mathematics
Kian Greene ('24) ERAU - Astronomy & Astrophysics and Computational Mathematics

Description: Pacific Northwest National Laboratory is enthusiastic to collaborate with students participating in the Embry-Riddle Aeronautical University Research Experience for Undergraduates (REU) on a research project focused on developing novel analytics approaches to analyzing signatures of nuclear materials. Optical spectroscopy is an approach that is used to perform real-time monitoring of nuclear materials production systems, and, in this project, we will partner to develop new computational techniques for characterizing varying levels of plutonium in different concentrations of nitric acid. Principal component analysis (PCA) is a method that is currently used, and the primary goals are to extend current PCA results to non-negative matrix factorization, which is tailored specifically to non-negative mixtures of elements.

Major outcomes:

  • Results were presented at 2022 ERAU REU Showcase
  • Publication is in preparation

Students

Emily Armstrong ('23)
Assumption University
Mathematics
Jensen Bridges ('24)
Oklahoma State University
Mathematics and Statistics
Zoe Friedman ('23)
Illinois State University
Mathematics
Jacob Antici ('24)
Arkansas State University
Mathematics
Max Prilutsky (‘22)
San Diego State University
Applied Mathematics
Zachariah Kline ('23)
Wisconsin Lutheran College
Mathematics and Computer Science
Emily Diegel ('24)
ERAU
Computational Mathematic
Rhiannon Hicks ('24)
ERAU
Astronomy & Astrophysics
Rachel Swan ('25)
ERAU
Computational Mathematic
Kian Greene ('24)
ERAU
Computational Mathematic + Astronomy & Astrophysics

CONTACT INFORMATION

Please contact REU site coordinator Dr. Berezovski at berezovm@erau.edu for any inquiry about REU Site and application process.