Active Few-Shot Learning for Vertex Classification Starting from an Unlabeled Dataset

2025-04-25

AI Summary

This paper addresses the challenge of prohibitively expensive vertex labeling and extreme scarcity of class labels in graph-structured data. We propose the first active few-shot vertex classification framework designed for the “no class oracle” setting—where neither class identities nor the number of classes are known a priori. Methodologically, we systematically decouple few-shot learning from class prior assumptions and introduce a three-stage active learning paradigm: (1) balanced sampling, (2) imbalanced cluster-based sampling, and (3) adaptation to unknown numbers of classes. Our approach integrates k-medoids clustering, prototypical networks (ProtoNet), and graph convolutional networks (GCN) to jointly optimize label recommendation and model training. Experiments show that ProtoNet consistently outperforms GCN under <20 labeled samples per class. When the class oracle is removed, GCN’s accuracy drops by 9%, whereas ProtoNet degrades by only 1%. Under unknown class counts, both models incur an additional ~1% performance loss, demonstrating ProtoNet’s superior robustness and practicality in realistic, label-scarce graph learning scenarios.

Technology Category

• Machine Learning: Active 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: Normalization, clustering, classification, and summarization of Web text

Abstract

Despite the ample availability of graph data, obtaining vertex labels is a tedious and expensive task. Therefore, it is desirable to learn from a few labeled vertices only. Existing few-shot learners assume a class oracle, which provides labeled vertices for a desired class. However, such an oracle is not available in a real-world setting, i.e., when drawing a vertex for labeling it is unknown to which class the vertex belongs. Few-shot learners are often combined with prototypical networks, while classical semi-supervised vertex classification uses discriminative models, e.g., Graph Convolutional Networks (GCN). In this paper, we train our models by iteratively prompting a human annotator with vertices to annotate. We perform three experiments where we continually relax our assumptions. First, we assume a class oracle, i.e., the human annotator is provided with an equal number of vertices to label for each class. We denote this as"Balanced Sampling''. In the subsequent experiment,"Unbalanced Sampling,'' we replace the class oracle with $k$-medoids clustering and draw vertices to label from the clusters. In the last experiment, the"Unknown Number of Classes,'' we no longer assumed we knew the number and distribution of classes. Our results show that prototypical models outperform discriminative models in all experiments when fewer than $20$ samples per class are available. While dropping the assumption of the class oracle for the"Unbalanced Sampling'' experiment reduces the performance of the GCN by $9%$, the prototypical network loses only $1%$ on average. For the"Unknown Number of Classes'' experiment, the average performance for both models decreased further by $1%$. Source code: https://github.com/Ximsa/2023-felix-ma

Problem

• Active few-shot learning for vertex classification without labeled data
• Overcoming lack of class oracle in real-world vertex labeling
• Comparing prototypical and discriminative models under varying annotation constraints

Innovation

• Active few-shot learning for vertex classification
• Iterative human annotation with balanced sampling
• Prototypical networks outperform discriminative models