Mini symposia/Special sessions

MS16 (TC304)Machine Learning and Data Analytics in Geotechnical Site Characterization

Associate Prof. Yu Wang: yuwang@cityu.edu.hk

Session Chairs:
Yu Wang, Associate Professor, City University of Hong Kong, E-mail: yuwang@cityu.edu.hk
Hui Wang, Assistant Professor, University of Dayton, USA, E-mail: hwang12@udayton.edu

Abstract of the special session:
Soils and rocks are natural geo-materials, and they are affected by many spatially varying factors during their complex geological formation process. The occurrence of geo-materials and their properties therefore vary spatially. Furthermore, such geo-materials are often underground and invisible at ground surface. Subsurface site characterization is therefore necessary before construction of any geotechnical project can commence. However, due to the difficulty in obtaining access to subsurface geo-materials and the constraints of investigation costs, manpower, and time, it is impossible to test subsurface geo-materials at every location within a site; only a small portion of geo-materials at some pre-selected locations can be examined during site characterization. This necessitates a planning of site characterization and effective analytics of the sparse and incomplete data from a specific site for best usage of the information obtained and risk-informed decision making. This problem becomes even more complicated when various uncertainties in site characterization are considered, such as spatial variability, transformation uncertainty, and statistical uncertainty. On the other hand, archived historical site investigation data and documents, if available, provide valuable data and information regarding subsurface configuration at the project site. These data, together with additional site investigation operations, become a “gold mine” for machine learning technics and data analytics to extract possible spatial and statistical patterns with quantified uncertainty. The topics relevant to this session include, but are not limited to, the following:

  • Machine learning methods
  • Sampling strategy (e.g., determination of sample size and locations)
  • Soil/rock databases and data management
  • Statistical characterization of soil/rock data
  • Correlations and transformation models
  • Characterization of spatial variability
  • Statistical and spatial uncertainty quantification in site characterization
  • Bayesian approaches
  • Value of information from site characterization
  • Impact on reliability‐based design
  • Information theory in geotechnics and engineering geology
  • Data interpretation

ICOSSAR 2021 Secretariat

Tongji University, 1239 Siping Road, Shanghai 200092, China    Email: icossar2021@tongji.edu.cn

© 2021 ICOSSAR 2021   Powered by Weicheng