Hierarchical Federated Learning with Dynamic Clustering and Adaptive Regularization for Robust Infrastructure Inspection
Researchers have developed a new hierarchical federated learning framework to address the challenges of data privacy and heterogeneity in infrastructure inspection using computer vision. The system uses dynamic clustering to group clients based on structural degradation and an adaptive regularization module to manage statistical imbalances within local datasets. This approach aims to create robust diagnostic models for complex infrastructure by overcoming dual-level heterogeneity without relying on geographical metadata. AI
IMPACT Introduces a novel framework for privacy-preserving collaborative learning in specialized AI applications like infrastructure inspection.