The dynamic spread of touch screen devices introduces a new dimension of user-system adaptation. The on-screen, user-generated touch events propose an unexplored portray of user behaviour. Here, the thesis focuses on obtaining the mental state, because this information enables the intense adaptation of the running software.
To be able to process the large amount of user generated data, it is recommended to build some kind of artificial intelligence into our system in order to deliver the mental state estimation needed for the adaptation.
In my thesis, I aim to solve the above mentioned classification problem by the application of deep neural networks. Neural network is a type of adaptive systems, used to solve complicated calculation problems with parallel processing. I introduce their structure, methods and application in my thesis.
Furthermore, I propose the algorithm design used for the evaluation of touch events with the estimation of the heart rate variability, and its connection with the mental workload.
The goal of my thesis is to help to transform the Android-based AdaptEd project’s educational games (developed on the university Department of Automation and Applied Informatics) to adapt to the touch events, the reaction time and the heart rate variability.