At the summer programme of the Institute of Oceanography, National Taiwan University (IONTU), Dr Lauriane Ribas-Deulofeu (Nord University) taught a course on AI applications in marine science.
Beyond the theoretical classes, the course included a two-hour active learning session. Groups worked through real case studies from BioBoost+, GuardIAS, and other projects. The question was never how to run an AI model. It was how to design an AI-assisted project: how to build a training set that gives robust and applicable results, how to minimise bias and keep both training sets and models transferable, and how to evaluate and report AI outputs transparently.
The session closed with an R tutorial on evaluating AI performance: confusion matrices, precision, recall, and what those numbers mean for the quality and robustness of outputs, and for their use by environmental managers and policy makers.
Attendees ranged from master’s students to full professors, which made for a diverse and sharp discussion, and for questions that will feed straight back into both projects.
Thanks to IONTU for hosting.
