The Partnership for Observation of the Global Oceans (POGO) 2019 Workshop on Machine Learning and Artificial Intelligence in Biological Oceanographic Observations took place at the Flanders Marine Institue in May of 2019. The workshop broadly aimed to introduce ocean scientists interested in sustained biological observations to modern machine learning tools. A total of 41 attendees representing 31 global institutions participated in a series of discussions and hands-on tutorials.
Linked to this page are the tutorial materials produced by the workshop organizers. The code and related text are divided by domain area: acoustic, genomics, and imaging. Each section consists of examples and exercises written in Python or R. The domain specific materials are independent and can be treated as seperate units. Likewise, each section has different dependencies and installation instructions.
Workshop attendees had access to computing resources courtesy of Amazon Web Services. With AWS, participants had access to powerful remote computing instances. Please note that certain examples might run quite slowly without the appropriate hardware support.
The acoustics tutorial was lead by Danelle Cline and John Ryan of the Monterey Bay Aquarium Research Institute. The Python based materials focus on passive acoustic monitoring of cetaceans with deep learning methods
Tristan Cordier of the University of Geneva and Anders Lanzén of AZTI-Tecnalia organized the genomics tutorial. Machine learning techniques for working with metabarcoding and metagenomic data are illustrated using code written in R. Techniques covered include logisitic regression, random forests, and support vector machines.
Eric Orenstein of the Scripps Institution of Oceanography and Simon-Martin Schröer of Kiel University produced a series of Jupyter notebooks written in Python. In situ images of marine plankton were used to illustrate a variety of concepts from region finding to classification with deep learning.