Engineering topics
Ries, Daniel
Publications and source records attributed to Ries, Daniel.
Assessing the reliability of prediction intervals from Bayesian Neural Networks.
Abstract not provided.
Rim-to-Rim Werables at The Canyon for Health (R2R WATCH): Physiological Cognitive and Biological Markers of Performance Decline in an Extreme Environment.
Abstract not provided.
Event Detection for Undersampled Electron Microscopy Experiments: A Control Chart Case Study
Real-time systems often consist of high-volume, high-velocity data streams in which making efficient and timely inferences about the data are challenging. In particular, when a user interacts with a real-time system to gain insights, detect events, and make decisions about the system, the rate and amount of information the user is required to process is generally overwhelming. In addition, analytically processing large volumes of data can be computationally expensive and, in real-time, renders traditional inferential methods effectively useless. One approach to mitigate these challenges is to reduce both the amount of information presented to the user and the volume of data placed in the stream. In this manuscript we will describe a method constructed under this approach in addition to the development and deployment of an online tool, Real time Event Detector for Subsampled Images (REDSI), designed to provide feedback on a real-time system by characterizing and detecting events of interest. We will discuss REDSI in the context of scanning transmission electron microscopy (STEM), which is a powerful real-time system that provides high spatial and temporal resolution on nanoscale structures and processes. The data produced by STEM experiments are a stream of images relaying structural, compositional, and dynamic interphase information to scientists in fields ranging from microbiology and neuroscience to materials science and energetics. While REDSI was designed to processes STEM images in real time, we will also discuss other real-time system applications for this methodology.
Using Bayesian Neural Networks for UQ of Hyperspectral Image Target Detection.
Abstract not provided.
Utilizing Distributional Measurements of Material Characteristics from SEM Images for Inverse Prediction.
Abstract not provided.
Using Bayesian Neural Network for UQ of Hyperspectral Image Target Detection.
Abstract not provided.
Modeling energy balance while correcting for measurement error via free knot splines
Measurements of energy balance components (energy intake, energy expenditure, changes in energy stores) are often plagued with measurement error. Doubly-labeled water can measure energy intake (EI) with negligible error, but is expensive and cumbersome. An alternative approach that is gaining popularity is to use the energy balance principle, by measuring energy expenditure (EE) and change in energy stores (ES) and then back-calculate EI. Gold standard methods for EE and ES exist and are known to give accurate measurements, albeit at a high cost. We propose a joint statistical model to assess the measurement error in cheaper, non-intrusive measures of EE and ES. We let the unknown true EE and ES for individuals be latent variables, and model them using a bivariate distribution. We try both a bivariate Normal as well as a Dirichlet Process Mixture Model, and compare the results via simulation. Our approach, is the first to account for the dependencies that exist in individuals’ daily EE and ES. We employ semiparametric regression with free knot splines for measurements with error, and linear components for error free covariates. We adopt a Bayesian approach to estimation and inference and use Reversible Jump Markov Chain Monte Carlo to generate draws from the posterior distribution. Based on the semiparameteric regression, we develop a calibration equation that adjusts a cheaper, less reliable estimate, closer to the true value. Along with this calibrated value, our method also gives credible intervals to assess uncertainty. A simulation study shows our calibration helps produce a more accurate estimate. Furthermore, our approach compares favorably in terms of prediction to other commonly used models.
Statistical Support for Experimental Design/Planning for Plutonium Signatures Project.
Abstract not provided.
A Geometric Approach for Calibrating Computer Models with Misaligned Functional Output.
Abstract not provided.
An Expert Test Bed for Nuclear Forensics Group Inclusion/Exclusion.
Abstract not provided.
Utilizing Distributional Measurements of Material Characteristics from SEM Images for Inverse Prediction.
Abstract not provided.