Animal Behavior & CognitionbioRxivPreprint
Classifier architecture and data preprocessing jointly shape accelerometer-based behavioural inference
Loïc Brun, Jonas Mateo Brady Rothrock, Erica van de Waal, Ebi Antony George
This study shows that deep learning models outperform classical methods in behavioural inference from accelerometers, but preprocessing choices significantly affect behaviour-specific accuracy.
1 citation