A joint research team of Osaka University and Nagoya City University has developed an artificial intelligence technology that understands the characteristics of behavior only from the movement trajectory of animals.It is expected that the behavior of various animals from nematodes to penguins can be correctly estimated with a probability of 90% or more, and that important brain activity can be found as well as the discovery of nests and feeding grounds.
According to Osaka University, the joint research team includes Professor Kotaro Kimura of the Graduate School of Information Science and Technology, Nagoya City University, Associate Professor Takuya Maekawa of the Graduate School of Information Science and Technology, Osaka University, and Shuhei Yamazaki, a specially appointed researcher of the Graduate School of Information Science and Technology, Osaka University. And others participated.
Until now, estimating the behavioral state of animals required sufficient knowledge and extremely large-scale data, but the collaborative research team combined two types of artificial intelligence technologies from only the movement trajectory of animals. We have developed an analysis method that estimates behavioral states and comprehensively extracts features.With this method, it is possible to estimate behavioral information with a relatively small number of data, such as dozens of animals.
A nematode that moves about 10 cm in 1 minutes in a petri dish and a penguin that swims several kilometers a day in the Antarctic Ocean can be analyzed in exactly the same way, and can be estimated correctly with a high probability of 1% or more.Furthermore, we found changes in neural activity related to behavioral changes from nematodes in the laboratory, and confirmed changes in behaviors caused by learning experiences from Drosophila and bats.
The collaborative research team believes that this method can be used as a behavior prediction when watching over elderly people and young children, in addition to improving understanding of animal behavior and finding nests.
Paper information:[Frontiers in Neuroscience] STEFTR: A hybrid versatile method for state estimation and feature extraction from the trajectory of animal behavior