Recognition and Enhancement of College Students' Self-Directed Learning Behaviors Based on Multilayer Perceptual Machines

Abstract

This paper proposes a method for identifying college students' autonomous learning behaviors based on a multi-layer perceptron. The process of identifying and analyzing college students' autonomous learning behaviors is constructed from four aspects: data collection, data preprocessing, data analysis, and online learning outcomes. Based on the characteristics of college students' online autonomous learning behaviors, the multi-layer perceptron method is employed to assess the performance of their autonomous learning behaviors. Learning behavior variables are selected from three dimensions: behavioral engagement, social engagement, and cognitive engagement. The impact of each learning behavior variable on learning outcomes is analyzed. Four key factors are identified: video viewing count, assignment practice count, unit test count, and participation count. Based on the performance levels of these four key factors, different learning modes are proposed, and the self-directed learning behavior performance of students in each learning mode is analyzed. The correlation coefficient between behavioral engagement and online learning outcomes is 0.724, indicating a high correlation. Among the four learning modes, “rhythmic-type” learners maintained stable learning efforts and good unit test scores throughout the course, with stable autonomous learning behavior performance and outstanding autonomous learning behavior capabilities. Enhancing college students' learning behavior can cultivate students to develop “rhythmic-type” learning behavior.