Statistical Machine Learning for Engineering with Applications

2025-01-30

Lecture Notes in Statistics

AI Summary

Accurate detection of fine, low-contrast, and non-uniform cracks in 3D concrete imagery remains challenging under high noise and limited labeled data. Method: This paper proposes an interpretability-driven statistical-learning coupled framework that uniquely embeds structural physics priors into the deep learning objective function, integrating Bayesian inference, regularized kernel methods, physics-informed neural networks (PINNs), and uncertainty quantification to establish a causally interpretable joint modeling paradigm. Contribution/Results: The approach significantly enhances model trustworthiness and cross-scenario generalizability, reducing prediction error by 32% across multiple engineering benchmarks. It enables real-time decision support while satisfying verifiability requirements stipulated by ASME and ISO standards.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Low Level & Physics-based Vision
• Reasoning under Uncertainty: Probabilistic Inference

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Problem

• 3D Concrete Imaging
• Crack Detection
• Machine Learning

Innovation

• RieszNet
• Generative Semi-Realistic Images
• Concrete Crack Detection