Graph Synthetic Out-of-Distribution Exposure with Large Language Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Xu, Haoyan, Yao, Zhengtao, Wang, Ziyi, Cheng, Zhan, Hu, Xiyang, Li, Mengyuan, Zhao, Yue
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910949850480640
author Xu, Haoyan
Yao, Zhengtao
Wang, Ziyi
Cheng, Zhan
Hu, Xiyang
Li, Mengyuan
Zhao, Yue
author_facet Xu, Haoyan
Yao, Zhengtao
Wang, Ziyi
Cheng, Zhan
Hu, Xiyang
Li, Mengyuan
Zhao, Yue
contents Out-of-distribution (OOD) detection in graphs is critical for ensuring model robustness in open-world and safety-sensitive applications. Existing graph OOD detection approaches typically train an in-distribution (ID) classifier on ID data alone, then apply post-hoc scoring to detect OOD instances. While OOD exposure - adding auxiliary OOD samples during training - can improve detection, current graph-based methods often assume access to real OOD nodes, which is often impractical or costly. In this paper, we present GOE-LLM, a framework that leverages Large Language Models (LLMs) to achieve OOD exposure on text-attributed graphs without using any real OOD nodes. GOE-LLM introduces two pipelines: (1) identifying pseudo-OOD nodes from the initially unlabeled graph using zero-shot LLM annotations, and (2) generating semantically informative synthetic OOD nodes via LLM-prompted text generation. These pseudo-OOD nodes are then used to regularize ID classifier training and enhance OOD detection awareness. Empirical results on multiple benchmarks show that GOE-LLM substantially outperforms state-of-the-art methods without OOD exposure, achieving up to a 23.5% improvement in AUROC for OOD detection, and attains performance on par with those relying on real OOD labels for exposure.
format Preprint
id arxiv_https___arxiv_org_abs_2504_21198
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Graph Synthetic Out-of-Distribution Exposure with Large Language Models
Xu, Haoyan
Yao, Zhengtao
Wang, Ziyi
Cheng, Zhan
Hu, Xiyang
Li, Mengyuan
Zhao, Yue
Machine Learning
Out-of-distribution (OOD) detection in graphs is critical for ensuring model robustness in open-world and safety-sensitive applications. Existing graph OOD detection approaches typically train an in-distribution (ID) classifier on ID data alone, then apply post-hoc scoring to detect OOD instances. While OOD exposure - adding auxiliary OOD samples during training - can improve detection, current graph-based methods often assume access to real OOD nodes, which is often impractical or costly. In this paper, we present GOE-LLM, a framework that leverages Large Language Models (LLMs) to achieve OOD exposure on text-attributed graphs without using any real OOD nodes. GOE-LLM introduces two pipelines: (1) identifying pseudo-OOD nodes from the initially unlabeled graph using zero-shot LLM annotations, and (2) generating semantically informative synthetic OOD nodes via LLM-prompted text generation. These pseudo-OOD nodes are then used to regularize ID classifier training and enhance OOD detection awareness. Empirical results on multiple benchmarks show that GOE-LLM substantially outperforms state-of-the-art methods without OOD exposure, achieving up to a 23.5% improvement in AUROC for OOD detection, and attains performance on par with those relying on real OOD labels for exposure.
title Graph Synthetic Out-of-Distribution Exposure with Large Language Models
topic Machine Learning
url https://arxiv.org/abs/2504.21198