In order to resolve the comprehension difficulties of theory and implementation about multi-objective decision in “Decision Analysis and Decision Support” course for postgraduates, digit recognition experiment is introduced into teaching practice. PCA method is used in the process of digit recognition, which is one of multi-objective decision methods. The digital recognition principles are firstly described based on PCA and BP neural network.

Then, a teaching experimental scheme is given, and the algorithm’s design and simulation can be carried in MATLAB. So that students can deeply understand and master the theoretical knowledge through the open experiment. It has been found that the experiment can effectively improve the students’ practical ability and the understanding of the process of multi-objective decision.