Journal of Higher Education Research

In order to further enhance the scientificity and adaptability of the value-added evaluation of college students' comprehensive quality, this study applies the machine learning algorithm model to realize the dynamic optimization of the weight on the basis of the initial weight determined by the analytic hierarchy process (AHP). By constructing a loss function with the goal of minimizing the value-added error of the comprehensive quality evaluation, the weight is used as a training parameter, and the gradient descent method is used to iteratively update the weight to realize the adaptive adjustment of the evaluation model. The dynamic weight optimization model can more accurately reflect the individual differences and group advantages of students' quality development, and can significantly improve the accuracy and feasibility of the value-added evaluation.