LEGO-Learn: Label-Efficient Graph Open-Set Learning

2024-10-21

arXiv.org

AI Summary

Addressing the dual challenges of low labeling cost and out-of-distribution (OOD) detection in open-set graph learning, this paper proposes a collaborative GNN framework. First, we design a GNN-based OOD filter jointly trained with a (C+1)-way classifier to simultaneously perform in-distribution (ID) classification and OOD detection. Second, we introduce a K-Medoids–based active sampling strategy to select highly informative ID nodes, thereby minimizing annotation effort. Third, we define a weighted cross-entropy loss that suppresses interference from OOD samples while strengthening supervision on critical ID samples. Extensive experiments on four real-world graph datasets demonstrate that our method achieves up to 6.62% improvement in ID classification accuracy and up to 7.49% gain in OOD detection AUROC, significantly outperforming existing low-budget open-graph learning approaches.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

How can we train graph-based models to recognize unseen classes while keeping labeling costs low? Graph open-set learning (GOL) and out-of-distribution (OOD) detection aim to address this challenge by training models that can accurately classify known, in-distribution (ID) classes while identifying and handling previously unseen classes during inference. It is critical for high-stakes, real-world applications where models frequently encounter unexpected data, including finance, security, and healthcare. However, current GOL methods assume access to many labeled ID samples, which is unrealistic for large-scale graphs due to high annotation costs. In this paper, we propose LEGO-Learn (Label-Efficient Graph Open-set Learning), a novel framework that tackles open-set node classification on graphs within a given label budget by selecting the most informative ID nodes. LEGO-Learn employs a GNN-based filter to identify and exclude potential OOD nodes and then select highly informative ID nodes for labeling using the K-Medoids algorithm. To prevent the filter from discarding valuable ID examples, we introduce a classifier that differentiates between the C known ID classes and an additional class representing OOD nodes (hence, a C+1 classifier). This classifier uses a weighted cross-entropy loss to balance the removal of OOD nodes while retaining informative ID nodes. Experimental results on four real-world datasets demonstrate that LEGO-Learn significantly outperforms leading methods, with up to a 6.62% improvement in ID classification accuracy and a 7.49% increase in AUROC for OOD detection.

Problem

• Train graph models to recognize unseen classes with minimal labeling
• Balance accurate ID classification and OOD detection under label constraints
• Reduce annotation costs while maintaining performance in real-world applications

Innovation

• GNN-based filter excludes potential OOD nodes
• K-Medoids algorithm selects informative ID nodes
• C+1 classifier balances OOD removal and ID retention