Horse recognition using inertial measurement units

In recent years, the use of equine data from Inertial Measurement Units(IMU's) has been going up in research as well as medicine. When this IMU data is used, it is often very useful to know from which horse it is. This paper shows a method that allows data from different horses to be separated, with the use of Artificial Neural Networks. This system was created with the use of a so-called Long Short-Term Memory Neural Network, also known as an LSTM. However, this method is reliant on the fact that there is data available from the horse that needs to be recognized. Because of this, this paper also proposes a method to recognize if data from a horse is not yet available, using a softmax probability baseline.