An event classification problem in collider physics is a fundamental problem to accelerate searches for new phenomena in nature. The event classification problem is aimed at discriminating the signal events of interest from the background events as much as possible. In this study, the transfer learning technique in deep learning was employed to efficiently address this event classification problem. In collider physics experiments, there are many types of data analyses that target different signal events. To ensure the transferability of these data analyses, a deep learning model based on a graph neural network architecture is proposed in this study. By applying the transfer learning with this model, we observed that a high accuracy of event classification can be achieved even with a small amount of data. In particular, a significant improvement was observed when the physics processes of the events were similar between the source and target datasets of the transfer learning. This achievement by the transfer learning provides a potential approach for saving computing resources for future collider experiments.