ROG$_{PL}$: Robust Open-Set Graph Learning via Region-Based Prototype Learning

TL;DR

This work tackles the challenge of open-set node classification on noisy graphs by introducing ROG_PL, a two-module framework that first denoises labels via similarity-based label propagation and then learns region-based open-set prototypes with interior and border components. The denoising stage refines pseudo-labels and filters confident samples, while the prototype stage builds a region-aware representation with a diversity-enforcing loss to reduce inter-class confusion and reserve capacity for unknown classes. The method delivers robust performance under both IND and OOD noise, outperforming a wide range of baselines on standard graph benchmarks and across near/far OOD scenarios, with additional support from ablation analyses. This approach offers practical benefits for real-world graph learning tasks where label noise and unseen categories are common, enabling more reliable detection of unknown classes while maintaining strong performance on known categories.

Abstract

Open-set graph learning is a practical task that aims to classify the known class nodes and to identify unknown class samples as unknowns. Conventional node classification methods usually perform unsatisfactorily in open-set scenarios due to the complex data they encounter, such as out-of-distribution (OOD) data and in-distribution (IND) noise. OOD data are samples that do not belong to any known classes. They are outliers if they occur in training (OOD noise), and open-set samples if they occur in testing. IND noise are training samples which are assigned incorrect labels. The existence of IND noise and OOD noise is prevalent, which usually cause the ambiguity problem, including the intra-class variety problem and the inter-class confusion problem. Thus, to explore robust open-set learning methods is necessary and difficult, and it becomes even more difficult for non-IID graph data.To this end, we propose a unified framework named ROG$_{PL}$ to achieve robust open-set learning on complex noisy graph data, by introducing prototype learning. In specific, ROG$_{PL}$ consists of two modules, i.e., denoising via label propagation and open-set prototype learning via regions. The first module corrects noisy labels through similarity-based label propagation and removes low-confidence samples, to solve the intra-class variety problem caused by noise. The second module learns open-set prototypes for each known class via non-overlapped regions and remains both interior and border prototypes to remedy the inter-class confusion problem.The two modules are iteratively updated under the constraints of classification loss and prototype diversity loss. To the best of our knowledge, the proposed ROG$_{PL}$ is the first robust open-set node classification method for graph data with complex noise.

Figures (2)

• Figure 1: Overview of the proposed ROG$_{PL}$. In the latent representation space, two modules are designed: denoising via label propagation and open-set prototype learning via regions. In specific, we first correct noisy labels through similarity-based label propagation and removes low-confidence samples, to solve the intra-class variety problem caused by noise. Then we learn open-set prototypes for each known class via non-overlapped regions and remains both interior and border prototypes to remedy the inter-class confusion problem. These two modules are iteratively updated under the constraints of classification loss and prototype diversity loss.
• Figure 2: The performance of ROG$_{PL}$ with respect to different IND noise rate and OOD noise rate on Cora and Citeseer datasets.