Improving Radiography Machine Learning Workflows via Metadata Management for Training Data Selection

TL;DR

This paper presents a metadata management tool for radiography machine learning workflows that enhances data selection, reduces redundancy, and improves reproducibility in scientific research pipelines.

Contribution

The paper introduces a novel metadata management tool tailored for dynamic radiography, demonstrating its effectiveness and discussing potential extensions to broader scientific machine learning workflows.

Findings

Improved data selection efficiency in radiography ML workflows

Enhanced reproducibility of machine learning experiments

Potential for reducing redundant work in scientific research pipelines

Abstract

Most machine learning models require many iterations of hyper-parameter tuning, feature engineering, and debugging to produce effective results. As machine learning models become more complicated, this pipeline becomes more difficult to manage effectively. In the physical sciences, there is an ever-increasing pool of metadata that is generated by the scientific research cycle. Tracking this metadata can reduce redundant work, improve reproducibility, and aid in the feature and training dataset engineering process. In this case study, we present a tool for machine learning metadata management in dynamic radiography. We evaluate the efficacy of this tool against the initial research workflow and discuss extensions to general machine learning pipelines in the physical sciences.