Abstract: This study designs a comprehensive molecular simulation experiment for predicting the antifouling performance of polymer materials using machine learning approaches. By compiling existing polymer structure-antifouling performance data from literature and calculating molecular descriptors through molecular simulations, we employ various machine learning methods—including genetic function approximation, neural network modeling, multiple linear regression, and partial least squares regression analysis—to establish quantitative structure-property relationship (QSPR) models for antifouling polymer materials. The predictive capabilities and goodness-of-fit of these models are systematically evaluated. This experiment enables students to acquire fundamental knowledge of machine learning principles and operational techniques, gain insights into recent advancements in marine antifouling materials research, and develop problem-solving skills in chemical applications through the integration of machine learning and molecular simulation methodologies.

Key words: Machine learning, Quantitative structure-property relationship (QSPR), Anti-fouling polymer materials, Molecular simulation