The objective of the workshop was to bring together the ISSMGE TC309 members and representatives from Norwegian institutions to discuss how various ML techniques are currently used in geoscience applications, and for a brainstorming of possible research collaborations on ML and big data in geoscience.
To facilitate understanding of the ML techniques and their implications in geoscience, ten presentation were delivered as case studies outlining experiences and lessons learned. The agenda can be seen here. Participants were encouraged to share experiences in challenges they faced and to discuss potential solutions to those issues.
The Machine learning workshop at NGI 22nd October 2018 gathered about 40 geotechnical engineers and IT experts.
Through the presentations and interactive break-out and discussions, many ideas emerged as potential areas that the TC309 (and geotechnical practitioners) could explore to push forward development of geotechnical applications of machine learning and help us make decisions. The key recommendations and possible actions include:
- Data Availability - A huge amount of monitoring and site investigation data (e.g. CPT data) are becoming available. We should take advantage of such data to improve our predictions across a wide range of geotechnical engineering applications. ML techniques can be used for to improve interpretation of geotechnical data in an automatic manner.
- Public Education Outreach - Education of geoscience professionals on ML techniques and artificial intelligence tool is a key issue in order to transfer from "black to grey box systems". Resistance within the community will decrease if practitioners have a working understanding of various Machine Learning techniques.
- Standards and Guidelines – Comply guidelines and recommendations to facilitate data collection and to ensure high quality datasets.
Zhongqiang Liu, senior engineer at NGI, is the chairman for the TC309, which was established by ISSMGE in 2018. TC309 aims to provide a forum for all interested members of ISSMGE to explore the use of machine learning (ML) techniques to solve problems relevant to soil mechanics and geotechnical engineering.