Summer Keynote Talk: Ramping up the AI Learning Curve: ‘Big Data’ applications to fisheries acoustics

At the June 11, 2024 American Fisheries Society Southern New England Chapter meeting, the keynote speaker will be Michael Jech from the NOAA/Northeast Fisheries Science Center. His talk title is, “Ramping up the AI Learning Curve: ‘Big Data’ applications to fisheries acoustics

Advances in active acoustic technology have outpaced our ability to process and analyze the data in a timely manner. Currently, scientists rely on manual scrutiny or limited automation to classify acoustic backscatter to biologically-meaningful taxa. Classification of acoustic backscatter to its source is a grand challenge of fisheries acoustics and many methods have been developed to partition backscatter for accurate abundance estimates of living marine resources. I explore artificial intelligence (AI), machine learning (ML), and other ‘Big Data’ methods to classify active acoustic data collected at the NOAA Northeast Fisheries Science Center. We have been collecting active acoustic data since 1998 and our processing and classification methods have evolved as we learn and as methods become available. Even with large volumes of data, we found we were limited by available training data for supervised approaches, so we are looking into unsupervised and self/semi-supervised methods as well as synthesizing training data using physics-based acoustic scattering models. These advanced analytical techniques are the next step in our quest to interpret remotely-collected data and provide accurate and timely information to stock assessment and ecosystem-based management.