Methodology for the Development of Adaptive Training Systems Based on Neural Network Methods

• Published in: Software Engineering Application in Informatics Vol. 232; pp. 238 - 253
• Main Authors: Obukhov, Artem, Vostrikova, Victoria, Nazarova, Alexandra, Teselkin, Daniil, Surkova, Ekaterina
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2021; Springer International Publishing
• Series: Lecture Notes in Networks and Systems
• Subjects: Adaptive training systems; Automation of the development of information systems; Machine learning; Methodology for the development of adaptive systems; Neural network methods
• ISBN: 9783030903176; 3030903176
• ISSN: 2367-3370; 2367-3389
• DOI: 10.1007/978-3-030-90318-3_23

The development of modern adaptive training systems (ATS) based on virtual reality technologies is an expensive process. To increase the efficiency of the development of such adaptive systems, a methodology is proposed that includes a set of neural network methods. This made it possible to automate the processes of analysis, information processing, decision-making, which reduced the software complexity of the ATS implementation and the cost. The article describes in detail the main stages of the methodology, the tools used in the course of its implementation. The methodology differs from existing approaches by the transition to neural network architecture and the addition of a stage for the implementation of neural network components. The proposed methodology has been successfully tested in the subject area of adaptive training systems. The application of the methodology and neural network methods in the implementation of the ATS for the training of miners made it possible to reduce economic costs by 18.9% and the complexity of software by 19.6%, to increase adaptability (by 20%), quality (by 12%) and productivity (by 5.5%) relative to the system developed according to RAD methodology and MVC architecture.