Challenges of machine learning model validation using correlated behaviour data: Evaluation of cross-validation strategies and accuracy measures.

Automated monitoring of the movements and behaviour of animals is a valuable research tool. Recently, machine learning tools were applied to many species to classify units of behaviour. For the monitoring of wild species, collecting enough data for training models might be problematic, thus we examine how machine learning models trained on one species can be applied to another closely related species with similar behavioural conformation. We contrast two ways to calculate accuracies, termed here as overall and threshold accuracy, because the field has yet to define solid standards for reporting and measuring classification performances. We measure 21 dogs and 7 wolves, and find that overall accuracies are between 51 and 60% for classifying 8 behaviours (lay, sit, stand, walk, trot, run, eat, drink) when training and testing data are from the same species and between 41 and 51% when training and testing is cross-species. We show that using data from dogs to predict the behaviour of wolves is feasible. We also show that optimising the model for overall accuracy leads to similar overall and threshold accuracies, while optimizing for threshold accuracy leads to threshold accuracies well above 80%, but yielding very low overall accuracies, often below the chance level. Moreover, we show that the most common method for dividing the data between training and testing data (random selection of test data) overestimates the accuracy of models when applied to data of new specimens. Consequently, we argue that for the most common goals of animal behaviour recognition overall accuracy should be the preferred metric. Considering, that often the goal is to collect movement data without other methods of observation, we argue that training data and testing data should be divided by individual and not randomly.

Companion and free-ranging Bali dogs: Environmental links with personality traits in an endemic dog population of South East Asia.

Dogs living on Bali Island have been free-ranging for thousands of years. A large group of expatriates sometimes adopt Bali dogs and keep them restricted to their houses and backyards, as is typical in modern western cultures. This provides us with the unique opportunity to compare the personality traits of dogs to their lifestyle either living as human companions or as free-ranging animals, exploring at the same time the impact of demographic variables (such as age, sex, and neutered status) on personality. After controlling for internal consistency of the scales and between-observer variation, we found that free-ranging Bali dogs were rated as less active, less excitable, less aggressive towards animals, and less inclined to chase animals or humans than Bali dogs living as human companions. Among free-ranging dogs, females were found to be more excitable. Females in the whole sample were also more fearful of people. The results of this preliminary study suggest that a change in lifestyle, i.e. being adopted, and living in a confined environment has negative consequences on some canine personality traits, such as activity/excitability, aggression towards animals, and prey drive.

Developmental exposure to xenoestrogen enhances spatial learning in male rats.

Steroid hormones have profound effects on the development and function of the nervous system. Environmental estrogens or xenoestrogens are manmade or are natural compounds, which mimics the action of estrogen hormones. The experimental evidence for impairment of cognitive functions in humans and mammals following exposure to xenoestrogens has been fiercely debated. The strongest arguments against such studies have been that the route, time course, and intensity of exposure did not simulate environmental exposure, and that the chemicals tested have additional, non-estrogenic toxic effects, hindering a generalization of actual "xenoestrogenic" effects. Here we show that an environmental-like exposure to the pure estrogen, 17alpha-ethynylestradiol (EE2) during development enhances spatial learning abilities in adult male Sprague-Dawley rats. To simulate an environmental exposure, we used a very low dose (4 ng/kg/day) of EE2 equivalent to concentrations measured in European and US streams which was given orally with a non-invasive method, and we extended the treatment for the entire course of development, from conception to puberty. The animals were tested in a Morris water maze protocol at 6 months of age. Male rats treated with EE2 during development showed a faster learning during the training phase, and remembered better the position of the hidden platform in the short term. Our study demonstrates that actual levels of exposure to xenoestrogens can permanently alter cognitive abilities of a mammalian species.