Tackling the Challenge of Small Sample Size: Effective Training by Integrating SNN and GNN

This complexity makes it more difficult to extract effective information and rules from small sample data. This article designs a method to address the challenge of small samples by integrating the advantages of SNN pulse neural network and GNN graph neural network. Using the time coding and threshold activation of SNN, it improves the excessive smoothing problem that exists in deep GNN. At the same time, it utilizes the advantages of GNN in image analysis to effectively capture and utilize complex relationships in structured data, and processes small sample graph structure data more effectively. Ultimately, it achieves the goal of effectively improving the difficulties of model training and generalization caused by small sample problems.