A Listening Ear in the Ocean: AI Decodes the Diet of Eagle Rays
Summary
Florida Atlantic University researchers have developed a machine learning-based acoustic monitoring system capable of detecting and classifying shell-crushing events from underwater feeding recordings of whitespotted eagle rays.
A New Way to Listen to the Seafloor
Florida Atlantic University researchers have developed a machine learning-based acoustic monitoring system capable of detecting and classifying shell-crushing events from underwater feeding recordings of whitespotted eagle rays. The technology enables remote measurement of predation pressure on mollusk populations at ecosystem scales.
The system demonstrated strong performance in field settings using animal-borne acoustic tags and fixed underwater recorders, even when trained exclusively on tank data.
How the System Works
The system uses a multi-step approach. First, it scans large datasets for potential shell-crushing sounds. Then, a second layer of machine learning is applied to reduce false detections.
After validation, the system classifies prey type using both traditional and deep learning methods, including:
- Random forests
- Long short-term memory networks
- Convolutional neural networks
Simpler Methods, Powerful Results
Interestingly, simpler methods using gammatone-based features performed nearly as well as advanced deep learning systems while requiring less computing power. This computational efficiency makes GTCC-based models suitable for autonomous underwater platforms with limited power and processing capacity.
Acoustic patterns reflected prey type, handling strategies, and processing time, potentially allowing distinction of individual feeding behaviors and prey size classes.
Publication & Support
The study was published in Ecological Informatics.
The research was supported by the Specialty License Plate fund administered by the Harbor Branch Oceanographic Institute Foundation and a National Science Foundation grant.