AttriReBoost: A Gradient-Free Propagation Optimization Method for Cold Start Mitigation in Attribute Missing Graphs

TL;DR

AttriReBoost (ARB) targets the cold-start problem in attribute-missing graphs by redefining boundary conditions and introducing virtual edges to boost gradient-free feature propagation. The method yields a provable convergence guarantee via a Banach fixed-point framework and retains low computational overhead comparable to propagation alone. Empirically, ARB achieves notable improvements in attribute reconstruction and downstream node classification across eight real-world datasets, while scaling to large graphs (e.g., 2.49M nodes on one GPU) with fast convergence. These properties make ARB a practical, scalable precursor or stand-alone component for graph-based learning with incomplete node attributes.

Abstract

Missing attribute issues are prevalent in the graph learning, leading to biased outcomes in Graph Neural Networks (GNNs). Existing methods that rely on feature propagation are prone to cold start problem, particularly when dealing with attribute resetting and low-degree nodes, which hinder effective propagation and convergence. To address these challenges, we propose AttriReBoost (ARB), a novel method that incorporates propagation-based method to mitigate cold start problems in attribute-missing graphs. ARB enhances global feature propagation by redefining initial boundary conditions and strategically integrating virtual edges, thereby improving node connectivity and ensuring more stable and efficient convergence. This method facilitates gradient-free attribute reconstruction with lower computational overhead. The proposed method is theoretically grounded, with its convergence rigorously established. Extensive experiments on several real-world benchmark datasets demonstrate the effectiveness of ARB, achieving an average accuracy improvement of 5.11% over state-of-the-art methods. Additionally, ARB exhibits remarkable computational efficiency, processing a large-scale graph with 2.49 million nodes in just 16 seconds on a single GPU. Our code is available at https://github.com/limengran98/ARB.

Figures (7)

• Figure 1: (a) Cold start examples in attribute-missing graphs. (b) Real-world datasets present a long-tail distribution, and tail/cold start nodes are difficult to participate in propagation.
• Figure 2: The overall framework of ARB. ARB address the cold start problem in attribute-missing graph learning by enhancing node connectivity through virtual edges and redefining initial boundary conditions. ARB boosts propagation-based methods to accurately and effectively recover missing attributes.
• Figure 3: Specific implementation process of attribute reconstruction and node classification.
• Figure 4: Training process and convergence speed.
• Figure 5: Hyperparameter value $\alpha$ and $\beta$ validation.