Machine Learning for Geoscience Data
A hands-on day on framing geoscience questions as supervised learning problems, from labelling through to honest validation.
Date
17 September 2026
Time
9:30 am – 4:00 pm AWST
Location
CIDS Training Lab, Building 105, Curtin University, Kent Street, Bentley WA 6102
Who it is for
Researchers, HDR students and industry analysts
Cost
Free for Curtin staff and students · $180 external
Register for this event
17 September 2026, 9:30 am – 4:00 pm AWST · CIDS Training Lab, Building 105. Free for Curtin staff and students · $180 external.
About this event
What happens on the day
Geoscience datasets rarely arrive model-ready. This workshop walks through the parts that decide whether a model is useful: how the target is defined, how the labels were produced, how spatial structure leaks between training and test splits, and how much of an accuracy number is real.
The day is built around a worked drill-and-blast dataset. You will move from raw survey exports to a trained classifier, then spend the afternoon breaking it, swapping the split strategy, removing features, and watching what happens to the metrics.
Bring a laptop. Environments are provided, so no local setup is needed. If you have a dataset of your own, bring it to the last session and we can talk about whether the approach transfers.
CIDS Training Lab, Building 105
Curtin University, Kent Street, Bentley WA 6102