ДОСЛІДЖЕННЯ ЕФЕКТИВНОСТІ ІДЕНТИФІКАЦІЇ ОСОБИ ЗА КЛАВІАТУРНИМ ПОЧЕРКОМ З УРАХУВАННЯМ СИЛИ ТИСКУ НА КЛАВІШІ

Using the "Queen Mary University Keystroke benchmark dataset" database and the Orange software, a study of the influence of time parameters, dynamics of changes in key pressure and their combinations on the accuracy of identification by keyboard handwriting was carried out. It has been experimentally confirmed that both for multi-class classification problems and for two-class classification problems based on key pressure signs, it is possible to obtain an identification accuracy 99 %.