STORY / PUBLIC SOURCE RECORD
Marine science: tremendous potential for machine learning
- Published 27 March 2023; updated 21 September 2026; estimated reading time 3 minutes. The article introduces machine learning applications at the Institute of Marine Research (IMR), based on work discussed by researchers Nils Olav Handegard, Ketil Malde and Howard Browman. It says machine learning can help find patterns in large, difficult-to-handle marine datasets and increase the scope of research.
- IMR researchers applied it to otoliths and fish scales, which can indicate fish age and life history in a way compared to tree rings. The article says performance can match human analysts in some cases; it also highlights narrower applications such as filtering irrelevant data and reducing repetitive work so experts can focus on analysis. A significant practical challenge is the time and effort needed to collect, organise and label training data, alongside the need to deploy tools consistently in research and advisory work.
- Properly trained algorithms can analyse large historical acoustic datasets and reveal patterns difficult to find manually. The article links this to increased computing power and lower costs for building large datasets, while describing wider adoption as an expectation.
- References: Malde, Handegard, Eikvil and Salberg, “Machine intelligence and the data-driven future of marine science,” ICES Journal of Marine Science 77(4), 2020, DOI 10.1093/icesjms/fsz057; Beyan and Browman, “Setting the stage for the machine intelligence era in marine science,” same issue, DOI 10.1093/icesjms/fsaa084. Original article published on IMR’s website 12 August 2020, credited to Liv Eva. Related industries: Aquaculture and fisheries; Technology. No direct contact block shown.