OD53A:
Machine Learning in Biological Oceanography II


Session ID#: 36969

Session Description:
Recent technological advances in instrumentation and computing have allowed scientists across all disciplines to collect an unprecedented amount of data. Biological oceanographers in particular are now faced with vast datasets that stymie traditional analysis methods. Scientists are increasingly leveraging machine learning (ML) techniques to process and analyze these information rich datasets. ML algorithms are designed to learn from one dataset to make accurate predictions about a new, independent one. While specific application domains might be quite different, the ML approaches used for analysis are often very similar. This session therefore aims to (1) identify new ML methods or applications, (2) examine overlap in disciplines applying similar ML techniques, (3) facilitate discussion and interdisciplinary collaborations among ML practitioners in the ocean science community, and (4) identify gaps and specific needs for oceanographers using ML. The session chairs welcome any submission detailing work on ML methods for ecological data analysis and inference in aquatic systems. The session is intended to have a broad scope and we invite abstracts from diverse fields such as imaging, acoustics, genomics, and modeling.
Primary Chair:  Eric Coughlin Orenstein, Scripps Institution of Oceanography, La Jolla, CA, United States
Co-chairs:  Jessica Y. Luo, National Center for Atmospheric Research, Climate and Global Dynamics, Boulder, CO, United States, John Burns, University of Hawai'i, Hawai'i Institute of Marine Biology, Papaikou, HI, United States and Ludwig Houegnigan, Polytechnic University of Catalonia, Barcelona Tech, ETSEIB, Barcelona, Spain
Moderators:  Eric Coughlin Orenstein, Scripps Institution of Oceanography, La Jolla, CA, United States and Jessica Y. Luo, National Center for Atmospheric Research, Climate and Global Dynamics, Boulder, CO, United States
Student Paper Review Liaison:  Eric Coughlin Orenstein, Scripps Institution of Oceanography, La Jolla, CA, United States
Index Terms:

4858 Population dynamics and ecology [OCEANOGRAPHY: BIOLOGICAL AND CHEMICAL]
4894 Instruments, sensors, and techniques [OCEANOGRAPHY: BIOLOGICAL AND CHEMICAL]
9820 Techniques applicable in three or more fields [GENERAL OR MISCELLANEOUS]
Cross-Topics:
  • F - Fisheries
  • IS - Ocean Observatories, Instrumentation and Sensing Technologies

Abstracts Submitted to this Session:

SELECTED APPLICATIONS OF MACHINE AND DEEP LEARNING TECHNIQUES TO VISUAL AND ACOUSTIC ECOSYSTEM MONITORING (321062)
Ludwig Houegnigan1, Pooyan Safari2, Climent Nadeu2 and Francesc Moreno-Noguer3, (1)Polytechnic University of Catalonia, Barcelona Tech, ETSEIB, Barcelona, Spain, (2)Polytechnic University of Catalonia, TALP Research Centre, Barcelona, Spain, (3)Polytechnic University of Catalonia, IRI, Barcelona, Spain
Data-driven decomposition of ocean observatory echosounder time series for ecological insights (319739)
Wu-Jung Lee1, Valentina Staneva2, Bernease Herman2 and Aleksandr Aravkin3, (1)University of Washington, Applied Physics Laboratory, Seattle, WA, United States, (2)University of Washington, eScience Institute, Seattle, WA, United States, (3)University of Washington, Department of Applied Mathematics, WA, United States
One Size Fits All? Determining the Impact of Image Size on Deep Learning Classifications of Marine Zooplankton (321958)
Jeffrey S Ellen, University of California, San Diego, Department of Computer Science, La Jolla, CA, United States and Mark D Ohman, Scripps Institution of Oceanography, UC San Diego, La Jolla, CA, United States
Machine Learning Approaches for Automated Taxonomic Classification of Plankton Images from Imaging FlowCytobot (322765)
Joe Futrelle and Anya M Waite, Woods Hole Oceanographic Institution, Woods Hole, MA, United States

See more of: Ocean Data Management