Adaptive Fitts for Adaptive Interface

Abstract

Adaptive interface would enable Human
Computer Interaction apply machine learning to cope with
human carelessness (mistakes), understand user performance
level and provide an interaction interface accordingly. This study
tends to translate the theoretical issues of human task into
working model by investigating and implementing the predicting
equation of human psychomotor behavior to a rapid and aimed
movement, developed by Paul Fitt in 1954. The study finds
logarithmic speed-accuracy trade-off and predict user
performance in a common task “point-select” using common
input device mouse. The performance of user is visualized as an
evidence and this visualization make a valuable step toward
understanding the change required in user interface to make the
interface adaptive and consistent. It proposed a method of
calculating the amount of change required through learning; add
extension to the theory of machine intelligence and increase
knowledge of Fitts applicability in terms of machine learning.