Speed Reading Trainer Based on Mind Machine Interface and Machine Learning Algorithms

In this work, we consider creating an application on iOS platform to develop speed reading skills. The application uses connection, data retrieval from the mind-machine interface device (MMI device) and machine learning algorithms in order to construct recommendations on the complexity level of the exercises. The proposed approach is based on the hypothesis confirmed: in the process of doing exercises of the same complexity level, not only the time spent for the exercise, but also the average level of attention (concentration) decrease - while the number of correct answers in relation to the wrong ones is kept in the same range. To construct a model of recommendations on the complexity level of the exercises, the cluster analysis is used – the done exercise is estimated for the need to change the level of complexity. In the framework of this work, the mobile application is developed on iOS platform, on the basis of which the development of speed reading skills through interactive exercises is implemented, also a user is able to connect MMI devices to the application, as well as receiving recommendations after performing the exercises. There are four different exercises available in the application, each of which is aimed at developing several skills, such as peripheral vision, concentration and development of short-term memory. For each exercise, a Python model based on the cluster analysis is created, which then is converted and used in the application via CoreML technology introduced by Apple in 2017. The effectiveness of the developed recommendation model is confirmed experimentally: the specified indicators are achieved faster by 6.67% by users receiving recommendations, rather than by users learn without recommendations. The work contains 52 pages, 3 chapters, 29 figures, 8 tables.