It strongly depends on human perceptual abilities, leaving lots of room for human error and making efficient tacit knowledge transfer in training. But even more importantly, human analysis of behavior is prone to subjectivity. As highlighted by Anderson and Perona ( 2), it is first of all a laborious and tedious task, limiting the volumes of processed data, as well as the number of analyzed behaviors or behavioral variables. However, it has long been acknowledged that relying on human observation imposes severe limitations on behavioral data acquisition and analysis. The seminal book “Measuring Behavior: an Introductory Guide” by Martin and Bateson ( 1) provides an excellent introduction to this topic. Traditionally, it is done through direct observation, and involves carefully designed steps: choosing the behavioral categories to observe, defining them in precise terms (usually they can have types of either event or state), deciding on the type of measurement, sampling method, etc.
Measuring behavior is key to behavioral testing, as well as many other behavior-related research methods in ecology, neuroscience, veterinary science, psychology, and many more. This demonstrates potential use of such clustering approach for exploration prior to hypothesis forming and testing in behavioral research. Using an example of protocol for testing in a “Stranger Test”, we compare the discovered clusters against the C-BARQ owner-based questionnaire, which is commonly used for dog behavioral trait assessment, showing that our method separated well between dogs with higher C-BARQ scores for stranger fear, and those with lower scores. To this end, we propose a concrete method for grouping video trials of behavioral testing of animal individuals into clusters using a set of potentially relevant features. We aim to demonstrate that such patterns can be useful at exploratory stages of data analysis before forming specific hypotheses. This study explores the potential of unsupervised approaches such as clustering for the automated discovery of patterns in data which have potential behavioral meaning. Unsupervised methods are increasingly used, but are under-explored in the context of behavior studies and applied contexts such as behavioral testing of dogs. Machine learning techniques are increasingly applied to support researchers in this field, mostly in a supervised manner: for tracking animals, detecting land marks or recognizing actions.
Traditional methods of data analysis in animal behavior research are usually based on measuring behavior by manually coding a set of chosen behavioral parameters, which is naturally prone to human bias and error, and is also a tedious labor-intensive task. 2Department of Biotechnology, Vives University College, Ghent, Belgium.
1Information Systems Department, University of Haifa, Haifa, Israel.Tom Menaker 1, Joke Monteny 2, Lin Op de Beeck 2 and Anna Zamansky 1 *
0 kommentar(er)
