Machine learning prediction models for AFM virtual imaging system

Abstract Nowadays, virtual laboratories are widely used in education and training universities. Such labs do gain some effect teaching, assisting students to be familiar with the experimental steps. However, these systems tend relatively simple. There is room for improvement helping understand principles. This particularly evident teaching of atomic force microscopy. In order overcome shortcomings AFM laboratory, we present a imaging system lower-resolution contact mode. We restore core principle beam deflection method using unity3D development platform. Several machine learning techniques employed build an prediction model. Since no public dataset available task topographical maps, create first grating samples prediction. The result indicates that proposed map model best performance CatBoost. prove feasibility building ability visualize internal structures predict sample maps. work has important applications related 3D dynamic display scanning process user experience training. At same time, it can help users get preliminary understanding different types under AFM, providing new idea construction laboratories.