OPTIMIZING LARGE-SCALE NETWORK EFFICIENCY USING GRAPH NEURAL NETWORKS AND PRINCIPAL COMPONENT ANALYSIS FOR REDUNDANCY REDUCTION

This manuscript proposes an integrated framework that combines Principal Component Analysis (PCA), Graph Convolutional Networks (GCNs), and Reinforcement Learning (RL) to improve the scalability and accuracy of node ranking in large-scale networks. PCA reduces feature dimensionality to eliminate redundancy, GCNs capture complex topological relationships among nodes, and RL iteratively refines ranking strategies based on structural feedback. This cohesive approach ensures efficient computation, preserves network structure, and enhances decision-making in network analysis tasks.