Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification

TL;DR

This paper introduces Transductive Sharpening, a training objective modification that leverages unlabeled node predictions by reducing entropy, improving semi-supervised node classification performance without changing model architecture.

Contribution

It proposes a novel loss-based method called Transductive Sharpening that utilizes unlabeled predictions to enhance node classification accuracy.

Findings

Consistent performance improvements across multiple benchmarks.

No architectural changes needed for the backbone models.

Effectively leverages unlabeled predictions through entropy minimization.

Abstract

In the transductive setting, where the full graph is observed but node labels are only partially available, progress in semi-supervised node classification has largely focused on architectural innovation. In this paper, we revisit an orthogonal axis: the training objective. We start from a simple observation: transductive models produce predictions for every node during training, including nodes without labels. These unlabeled-node predictions may contain useful training signal, but standard supervised objectives discard them because no ground-truth labels are available. Inspired by the decomposition of cross-entropy into a label-dependent alignment term and a label-independent entropy term, we propose prediction confidence as a natural way to extract this signal in the absence of labels. This motivates Transductive Sharpening (TS): a loss-level modification that minimizes prediction entropy on unlabeled nodes while counterbalancing this effect on labeled nodes. We evaluate Transductive Sharpening across a wide range of node-classification benchmarks and observe consistent performance improvements without requiring any changes to the backbone architecture. Code is available at https://github.com/transductive-sharpening/tunedGNN.