Feedforward-Aided Course Designs for Similarity Search

In this paper, we present two feedforward-aided designs for a Master's level course on similarity search based on different teaching methods: In project-based learning , the students are encouraged to learn autonomously while working on non-trivial real-world problems. Students address a problem over several months by creating an artifact (e.g., by implementing an algorithm). A similar but different teaching method is task-based learning. Rather than working on long-lasting projects, students work on smaller (but useful) tasks. In both course designs, we employ an auto-grader to provide students with automated and instant feedforward in a continuous manner, which allows them to improve their performance autonomously. We discuss and share our experiences with applying both methods in class. Furthermore, we give insights on the course evaluation based on the student's feedback, share our lessons learned, and analyze the student's grades.