A FRAMEWORK FOR THE DESIGN, IMPLEMENTATION AND EVALUATION OF AI BASED REAL-LIFE LEARNING SCENARIOS FOR COMPUTER SCIENCE NON-MAJORS
University of Applied Sciences Neu-Ulm (GERMANY)
About this paper:
Conference name: 15th annual International Conference of Education, Research and Innovation
Dates: 7-9 November, 2022
Location: Seville, Spain
Abstract:
The accelerated development in the field of immersive technologies (e.g. machine learning) raises significant challenges in teaching. It is often difficult, especially for undergraduate non-major computer science students, to learn sustainably the fundamentals of computer science subjects in appropriate depth and breadth as well as to apply the knowledge. The approach in this paper attempts to deal with these challenges by describing an innovative learning scenario along with a suitable IT infrastructure.
In the course, the students design and implement a distributed application that allows for identifying and classifying traffic signs based on photos from real traffic situations using methods of computer vision and machine learning. By developing this application, the students learn about mobile app development using Java for front-end, back-end development for object detection with Python notebooks on JupyterHub as well as the communication between these components.
The learning scenario we describe in this paper is for a period of one semester (four months) and contains the overall structure, suitable IT infrastructure, timeline and milestones. Within this innovative approach, we combine blended learning elements for the teaching and examining of the theoretical fundamentals as well as a collaborative and sand-box oriented development environment with a framework of libraries and code examples.
Moreover, the described learning scenario is a collaborative approach that is completely remote capable. Consequently, students are able to work together within their preassigned groups even during pandemic situations while still be supervised adequately. After the course, the students kindly filled in questionnaires. The evaluations of the questionnaires demonstrated the positive impact not only in the purely technical dimensions but also in terms of increasing the students' motivation and commitment to learning. By including elements of the Intrinsic Motivation Inventory (IMI), a comparability to related work is given.
The innovative approach described here is transferable to other non-major computer science programs because of the overall design of the course: collaborative blended learning formats with client-server architecture for scalable group sizes even in distance learning situations and details about the IT infrastructure. Keywords:
Traffic sign recognition, object detection, machine learning, teaching scenario, Python.