Related Experiment Video
Updated: Sep 30, 2025
11:56
Training Rats to Voluntarily Dive Underwater: Investigations of the Mammalian Diving Response
Published on: November 12, 2014
12.5K
Deep inference of seabird dives from GPS-only records: Performance and generalization properties
Amédée Roy1,2, Sophie Lanco Bertrand1, Ronan Fablet2
1Institut de Recherche pour le Développement (IRD), MARBEC (Univ. Montpellier, Ifremer, CNRS, IRD), Sète, France.
Plos Computational Biology
|March 11, 2022
Summary
Deep learning models accurately predict seabird dives from GPS tracking data, outperforming traditional methods. These models generalize well across different colonies and species, offering a powerful tool for marine ecology research.
Area of Science:
- Marine Ecology
- Animal Behaviour
- Computational Biology
Background:
- Seabird foraging behaviour is crucial for understanding population dynamics and marine ecosystem health.
- GPS tracking provides valuable seabird trajectory data, but inferring behaviour requires advanced analytical methods.
- Deep learning shows promise for classifying animal behaviour from trajectory data, yet requires further investigation into optimal architectures and generalization.
Purpose of the Study:
- To benchmark deep neural network architectures for predicting seabird dives from GPS trajectory data.
- To assess the generalization capabilities of trained deep learning models across different seabird colonies and ecosystems.
- To explore cross-species generalization using transfer learning for behaviour prediction.
Main Methods:
- Supervised training of deep neural networks on a dataset of ~300 seabird foraging trajectories with simultaneous pressure sensor data.
- Benchmarking deep learning models against Hidden Markov Models for dive prediction accuracy.
- Evaluating model generalization using data from different colonies and species, including fine-tuning pre-trained convolutional networks.
Main Results:
- Deep learning models significantly outperform Hidden Markov Models in predicting seabird dives.
- Trained convolutional networks demonstrate strong generalization, accurately predicting dives for seabirds from different colonies and ecosystems.
- Transfer learning via fine-tuning reduced the required dataset size for accurate dive prediction in a new species.
Conclusions:
- Deep learning offers a superior approach for analyzing seabird dive behaviour from trajectory data.
- The developed deep learning models exhibit robust generalization properties, applicable to diverse seabird populations and environments.
- This study provides a foundation for future applications of fine-tuning deep learning models in seabird behavioural ecology.
Related Concept Videos
01:11Buoyancy and Stability for Submerged and Floating Bodies
2.1K
In fluid mechanics, buoyancy and stability are key concepts for understanding the behavior of submerged and floating bodies. When a stationary body is fully or partially submerged in a fluid, the fluid exerts a force on the body known as the buoyant force. This force acts vertically upward through a point called the center of buoyancy, which is the center of the displaced fluid volume. According to Archimedes' principle, the magnitude of the buoyant force is equal to the weight of the fluid...
2.1K
01:24Generalization, Discrimination, and Extinction
860
Generalization, discrimination, and extinction are key concepts in operant conditioning that influence how behaviors are learned and maintained.
Generalization occurs when a behavior reinforced in one context is performed in similar situations. For instance, a student who studies diligently for calculus and receives excellent grades might apply the same study habits to psychology and history, expecting similar results. Generalization shows how learning in one setting can influence behavior in...
Generalization occurs when a behavior reinforced in one context is performed in similar situations. For instance, a student who studies diligently for calculus and receives excellent grades might apply the same study habits to psychology and history, expecting similar results. Generalization shows how learning in one setting can influence behavior in...
860