Solving BM/AirportSoilProperties/2/2025 using hierarchical Bayesian model-based stacking ensemble learning framework
Year: 2026
Type: article
Abstract: This short communication solves the benchmarking problems BM/AirportSoilProperties/2/2025 using the HBM-based stacking ensemble learning framework (HBM-based SELF). The HBM-based SELF is based on a series of base learners, which are variants of hierarchical Bayesian model (HBM) with various degrees of complexity. The best learner is selected as the base learner with the least predictive entropy. Benchmark problems #1 and #2 (BM1 and BM2) in BM/AirportSoilProperties/2/2025 are both solved, and single-borehole and collective-borehole predictions in BM1 are both addressed. The prediction accuracy of the best learner selected by the HBM-based SELF is quantified by the root mean squared error (RMSE), and the computation runtime for the HBM-based SELF is reported. The relative frequency that each base learner is selected as the best learner is summarized.
Source: Geodata and AI.
Authors: Jianye Ching, Kuan-Hsiang Lo
Institution: National Taiwan University
Language: English
Subfield: Safety, Risk, Reliability and Quality
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,000
OpenAlex ID: W7130510236