Where LLM Annotators Fail: Label-Free Learning on Graphs with LLMs

• URL: http://arxiv.org/abs/2605.27913v1
• Date: Wed, 27 May 2026 03:41:04 GMT
• Title: Where LLM Annotators Fail: Label-Free Learning on Graphs with LLMs
• Abstract summary: Cluster-Aware Noise Estimation (CANE) is a label-free learning framework that estimates cluster-conditional LLM reliability without ground truth labels.<n>CANE improves over the strongest label-free baselines, with the largest gains on datasets exhibiting stronger cluster-conditional noise.
• Score: 25.577309437856428
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Node classification on graphs often requires labeled nodes, yet obtaining labels at graph scale is expensive. When node attributes contain semantic content, such as paper abstracts, web pages, or product descriptions, large language models (LLMs) can provide low-cost supervision by annotating a small subset of nodes. However, these LLM-generated labels are noisy, and existing label-free graph learning methods usually treat this noise as either global or class-conditional. We find that LLM annotation errors are not only class-dependent but also region-dependent: within the same class, reliability can vary sharply across feature-space clusters. In light of this, we propose Cluster-Aware Noise Estimation (CANE), a label-free learning framework that estimates cluster-conditional LLM reliability without ground truth labels, and uses this estimate to decide which pseudo-labels to trust, and which labels to correct. Across various graph benchmarks and GNN backbones, CANE improves over the strongest label-free baselines, with the largest gains on datasets exhibiting stronger cluster-conditional noise.

Related papers

• PaSta: Noisy Node Classification with Partial Label Learning [58.43127707101144]
  We propose a novel Partial label-based Self-training framework to overcome the limitations of existing methods.<n>PaSta achieves an average improvement of 1.1% in classification performance under various noise settings.
  arXiv  Detail & Related papers  (2026-08-26T04:40:23Z)
• HopRank: Self-Supervised LLM Preference-Tuning on Graphs for Few-Shot Node Classification [34.92914341967866]
  We present HopRank, a fully self-supervised framework for node classification on text-attributed graphs.<n>We show that HopRank matches fully-supervised GNNs and substantially outperforms prior graph-LLM methods.
  arXiv  Detail & Related papers  (2026-04-19T05:58:38Z)
• GNN-as-Judge: Unleashing the Power of LLMs for Graph Learning with GNN Feedback [29.323513184371222]
  Large Language Models (LLMs) have shown strong performance on text-attributed graphs (TAGs)<n>This paper targets two key challenges: (i) the difficulty of generating and selecting reliable pseudo labels on TAGs for LLMs, and (ii) the need to mitigate potential label noise when fine-tuning LLMs with pseudo labels.<n>We propose a new framework, GNN-as-Judge, which can unleash the power of LLMs for few-shot semi-supervised learning on TAGs.
  arXiv  Detail & Related papers  (2026-03-11T21:14:33Z)
• Semi-supervised Instruction Tuning for Large Language Models on Text-Attributed Graphs [62.544129365882014]
  We propose a novel Semi-supervised Instruction Tuning pipeline for Graph Learning, named SIT-Graph.<n> SIT-Graph is model-agnostic and can be seamlessly integrated into any graph instruction tuning method that utilizes LLMs as the predictor.<n>Extensive experiments demonstrate that when incorporated into state-of-the-art graph instruction tuning methods, SIT-Graph significantly enhances their performance on text-attributed graph benchmarks.
  arXiv  Detail & Related papers  (2026-01-19T08:10:53Z)
• When Noisy Labels Meet Class Imbalance on Graphs: A Graph Augmentation Method with LLM and Pseudo Label [3.667121386226796]
  This paper systematically investigates robust node classification for class-imbalanced graphs with noisy labels.<n>We propose GraphALP, a novel Graph Augmentation framework based on Large language models (LLMs) and Pseudo-labeling techniques.<n> Experimental results show that GraphALP achieves superior performance over state-of-the-art methods on class-imbalanced graphs with noisy labels.
  arXiv  Detail & Related papers  (2025-07-24T07:39:07Z)
• Calibrating Pre-trained Language Classifiers on LLM-generated Noisy Labels via Iterative Refinement [8.804897656598051]
  We propose SiDyP: Simplex Label Diffusion with Dynamic Prior to calibrate the classifier's prediction.<n>Our framework can increase the performance of the BERT classifier fine-tuned on both zero-shot and few-shot LLM-generated noisy label datasets by an average of 7.21% and 7.30% respectively.
  arXiv  Detail & Related papers  (2025-05-26T08:31:55Z)
• Dynamic Bundling with Large Language Models for Zero-Shot Inference on Text-Attributed Graphs [56.73437142790496]
  Large language models (LLMs) have been used in many zero-shot learning problems.<n>LLMs struggle with text attributes isolated from the graph topology.<n>They yield unreliable predictions due to both information insufficiency and the inherent weakness of LLMs.
  arXiv  Detail & Related papers  (2025-05-23T08:12:16Z)
• Leveraging Large Language Models for Effective Label-free Node Classification in Text-Attributed Graphs [10.538099379851198]
  Locle is an active self-training framework that does Label-free node Classification with LLMs cost-Effectively.<n>It iteratively identifies small sets of "critical" samples using GNNs and extracts informative pseudo-labels for them with both LLMs and GNNs.<n>It significantly outperforms state-of-the-art methods under the same query budget to LLMs in terms of label-free node classification.
  arXiv  Detail & Related papers  (2024-12-16T17:04:40Z)
• Inaccurate Label Distribution Learning with Dependency Noise [52.08553913094809]
  We introduce the Dependent Noise-based Inaccurate Label Distribution Learning (DN-ILDL) framework to tackle the challenges posed by noise in label distribution learning. We show that DN-ILDL effectively addresses the ILDL problem and outperforms existing LDL methods.
  arXiv  Detail & Related papers  (2024-05-26T07:58:07Z)
• ERASE: Error-Resilient Representation Learning on Graphs for Label Noise Tolerance [53.73316938815873]
  We propose a method called ERASE (Error-Resilient representation learning on graphs for lAbel noiSe tolerancE) to learn representations with error tolerance. ERASE combines prototype pseudo-labels with propagated denoised labels and updates representations with error resilience. Our method can outperform multiple baselines with clear margins in broad noise levels and enjoy great scalability.
  arXiv  Detail & Related papers  (2023-12-13T17:59:07Z)
• BadLabel: A Robust Perspective on Evaluating and Enhancing Label-noise Learning [113.8799653759137]
  We introduce a novel label noise type called BadLabel, which can significantly degrade the performance of existing LNL algorithms by a large margin. BadLabel is crafted based on the label-flipping attack against standard classification. We propose a robust LNL method that perturbs the labels in an adversarial manner at each epoch to make the loss values of clean and noisy labels again distinguishable.
  arXiv  Detail & Related papers  (2023-05-28T06:26:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.