Conference Tools for MS&T24: Materials Science & Technology

Login

Register as a New User

Help

Submit An Abstract

Propose A Symposium

Presenter/Author Tools

Organizer/Editor Tools

About this Abstract
Meeting	MS&T24: Materials Science & Technology
Symposium	Uncertainty Quantification Applications in Materials and Engineering
Presentation Title	Uncertainty Quantification in Machine Learning Models with High-Dimensional Features and Large Sample Size
Author(s)	Ayorinde Emmanuel Olatunde, Weiqi Yue, Roger H. French, Pawan Tripathi, Anirban Mondal
On-Site Speaker (Planned)	Ayorinde Emmanuel Olatunde
Abstract Scope	Uncertainties within machine learning (ML) models, which are broadly divided into aleatory (reducible) and epistemic (irreducible) categories, may arise at the input level, output level, or a combination of both. Despite the increasing recognition of the importance of UQ in Engineering and Science, UQ currently faces scalability challenges, particularly in dealing with high-dimensional features and large sample sizes. Our project aims to conduct UQ on the ML model (specifically, a Gaussian Process) employed in predicting the β-phase volume fraction of Ti–6Al–4V alloy during heat treatment. This prediction is based on features comprising high-dimensional image sequences of 2D diffraction patterns captured at a synchrotron beamline at a scaled level. This project is still in its early stages, providing ample room for exploration, especially as we seek to apply UQ to scenarios involving both high-dimensional features and large sample sizes. Consequently, the findings from our research may emerge gradually in different phases.

OTHER PAPERS PLANNED FOR THIS SYMPOSIUM

A Parametric Study of Optical Floating-Zone Crystal-Growth Furnace Through Modeling of Heat Transfer: Effect of Sample Properties and Environment Gas Pressure

Automating Engineering Design with UQ-Aware Scientific Learning

Bayesian Calibration of Cladding Creep Model Coefficients in the PAD5 Fuel Performance Code Using the Dakota Toolkit

Bayesian Protocols for High-Throughput Optimization of Kinematic Hardening Models Using Cyclic Microindentation Experiments

Introduction to Verification, Validation, and Uncertainty Quantification for Engineering Simulation

Quantification of Uncertainty in Microstructure Segmentation of Solid Oxide Cell Electrodes Using an Improved Watershed Methodology

Quantitative Analysis of Systematic Uncertainties in Empirical and Machine Learning Interatomic Potentials

Tasmanian Toolkit for Uncertainty Quantification

Uncertainty Quantification in Machine Learning Models with High-Dimensional Features and Large Sample Size

Uncertainty Quantification of Material Properties in Data-Poor Regimes Using Transfer Learning and Gaussian Process Regression

Unraveling Correlation between Interface Structure and Magnetic Properties of La1-xSrxCoO3−δ/La1-xSrxMnO3−δ Bilayers Using Neural Architecture Search and Deep Ensembles

Questions about ProgramMaster? Contact programming@programmaster.org