🤖 AI Summary
To address insufficient node feature expressiveness in graph node classification, this paper proposes Learnable Generalized Geodesic Distance (LGGD). LGGD is the first method to enable end-to-end differentiable learning of parameters in the generalized geodesic distance equation over graphs, jointly modeling graph topology, node content, and supervision signals to produce robust, enhanced node representations. A key innovation is a dynamic label injection mechanism that significantly improves robustness against label noise and outliers. Extensive experiments on multiple real-world graph benchmarks demonstrate that LGGD achieves state-of-the-art or superior performance in node classification, validating the effectiveness and generalizability of learnable geodesic geometric modeling for graph representation learning.
📝 Abstract
Geodesic distances on manifolds have numerous applications in image processing, computer graphics and computer vision. In this work, we introduce an approach called `LGGD' (Learned Generalized Geodesic Distances). This method involves generating node features by learning a generalized geodesic distance function through a training pipeline that incorporates training data, graph topology and the node content features. The strength of this method lies in the proven robustness of the generalized geodesic distances to noise and outliers. Our contributions encompass improved performance in node classification tasks, competitive results with state-of-the-art methods on real-world graph datasets, the demonstration of the learnability of parameters within the generalized geodesic equation on graph, and dynamic inclusion of new labels.