Unraveling the Impact of Heterophilic Structures on Graph Positive-Unlabeled Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Yuhao, Yao, Jiangchao, Han, Bo, Yao, Lina, Liu, Tongliang
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913373076062208
author Wu, Yuhao
Yao, Jiangchao
Han, Bo
Yao, Lina
Liu, Tongliang
author_facet Wu, Yuhao
Yao, Jiangchao
Han, Bo
Yao, Lina
Liu, Tongliang
contents While Positive-Unlabeled (PU) learning is vital in many real-world scenarios, its application to graph data still remains under-explored. We unveil that a critical challenge for PU learning on graph lies on the edge heterophily, which directly violates the irreducibility assumption for Class-Prior Estimation (class prior is essential for building PU learning algorithms) and degenerates the latent label inference on unlabeled nodes during classifier training. In response to this challenge, we introduce a new method, named Graph PU Learning with Label Propagation Loss (GPL). Specifically, GPL considers learning from PU nodes along with an intermediate heterophily reduction, which helps mitigate the negative impact of the heterophilic structure. We formulate this procedure as a bilevel optimization that reduces heterophily in the inner loop and efficiently learns a classifier in the outer loop. Extensive experiments across a variety of datasets have shown that GPL significantly outperforms baseline methods, confirming its effectiveness and superiority.
format Preprint
id arxiv_https___arxiv_org_abs_2405_19919
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Unraveling the Impact of Heterophilic Structures on Graph Positive-Unlabeled Learning
Wu, Yuhao
Yao, Jiangchao
Han, Bo
Yao, Lina
Liu, Tongliang
Machine Learning
Social and Information Networks
While Positive-Unlabeled (PU) learning is vital in many real-world scenarios, its application to graph data still remains under-explored. We unveil that a critical challenge for PU learning on graph lies on the edge heterophily, which directly violates the irreducibility assumption for Class-Prior Estimation (class prior is essential for building PU learning algorithms) and degenerates the latent label inference on unlabeled nodes during classifier training. In response to this challenge, we introduce a new method, named Graph PU Learning with Label Propagation Loss (GPL). Specifically, GPL considers learning from PU nodes along with an intermediate heterophily reduction, which helps mitigate the negative impact of the heterophilic structure. We formulate this procedure as a bilevel optimization that reduces heterophily in the inner loop and efficiently learns a classifier in the outer loop. Extensive experiments across a variety of datasets have shown that GPL significantly outperforms baseline methods, confirming its effectiveness and superiority.
title Unraveling the Impact of Heterophilic Structures on Graph Positive-Unlabeled Learning
topic Machine Learning
Social and Information Networks
url https://arxiv.org/abs/2405.19919