Reproduce “Bridge type classification: supervised learning on a modified NBI dataset” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Achyuthan Jootoo, David Lattanzi
Researchers at George Mason University developed supervised machine learning models to classify bridge structural types based on preliminary design parameters, enhancing the initial design phase by providing data-driven recommendations. Their state-specific classifiers, utilizing selected NBI attributes and seismic data, achieved an average recall of 88.6% and precision of 88.0%, demonstrating the potential for more optimized bridge designs.
Key insights
- Four NBI attributes (deck structure type, average span length, maximum span length, material type) were identified as statistically most relevant for bridge type classification.
- The inclusion of USGS seismic intensity data consistently improved model predictive accuracy, especially in seismically active states.
- State-specific models significantly outperformed a unified nationwide model, underscoring the importance of regional context for bridge design predictions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.