Development of a methodological framework for a robust prediction of the main behaviours of dairy cows using a combination of machine learning algorithms on accelerometer data
Lucile Riaboff, Sylvain Poggi, Aurélien Madouasse, Sébastien Couvreur, S. Aubin, Nicolas Bédère, E. Goumand, Alain Chauvin, Guy Plantier
Abstract
• Main behaviours of dairy cows were successfully predicted using accelerometer data. • EXtreme Gradient Boosting followed by the Viterbi algorithm led to the best results. • Postures are the most difficult to discriminate with an accelerometer on the neck. • 86 Holstein cows from 4 farms were equipped and observed leading to a large dataset. • Independent signal sequences with a stratification were used to validate the models.
Topics & Concepts
AccelerometerGradient boostingMachine learningViterbi algorithmAlgorithmBoosting (machine learning)Artificial intelligenceComputer sciencePattern recognition (psychology)Random forestHidden Markov modelOperating systemGenetic and phenotypic traits in livestockAnimal Behavior and Welfare StudiesEffects of Environmental Stressors on Livestock