Implicit Feature Reweighting for Label-Free Graph Generalization

Tianjun Yao; Chen Xi; Xinyang Li; Lixian Zhong

cs.LGOct 21, 2026

Implicit Feature Reweighting for Label-Free Graph Generalization

Authors: Tianjun Yao, Chen Xi, Xinyang Li, Lixian Zhong

Organizations: 1Shenzhen Loop Area Institute · 2Guangzhou University · 3North China Electric Power University · University of Science and Technology of China

Abstract

Graph-level out-of-distribution (OOD) generalization is critical for deploying graph neural networks in real-world applications where distribution shifts are inevitable. Existing approaches typically require class labels during representation learning to decide which substructures are predictive, limiting their applicability. We propose SCARCE, a label-free representation-learning framework that yields distribution-shift-robust graph representations from an environment partition of the training graphs alone, whether that partition comes from dataset metadata or, when none is available, from clustering a self-supervised embedding. Our method operates in two stages: (1) a variational graph autoencoder is warm-started for graph reconstruction; and (2) an adversarial confusion mechanism forces the encoder to produce representations from which a discriminator cannot distinguish environments, while preserving graph-reconstruction fidelity. We provide theoretical analysis showing that environment confusion contracts environment-discriminative directions in the representation space, which, under regularized downstream classifiers, is equivalent to imposing a stronger directional penalty on environment-associated features. This feature-reweighting mechanism reduces reliance on spurious correlations without erasing information, offering a label-free, continuous relaxation of last-layer reweighting methods. Experiments on a controlled synthetic datasets and the DrugOOD datasets show that SCARCE achieves state-of-the-art OOD generalization among label-free methods, using no class labels during representation learning.

Figures & tables

Figure 1: Overview of the Scarce framework. Given an environment partition of the training graphs (dataset metadata or, when unavailable, clusters of a frozen self-supervised embedding), (1) a VGAE is warm-started for graph reconstruction and (2) an adversarial confusion mechanism forces the encoder to produce representations indistinguishable across environments while preserving reconstruction fidelity.

Figure 3: Controlled contraction of an environment-associated direction. Cells report $\Delta R_{\mathrm{OOD}}$ (10 seeds; negative = improvement). (a) Contraction helps when environment and task directions are separated but hurts at full alignment. (b) The effect vanishes as $\lambda\to 0$ and peaks at intermediate $\lambda$ , consistent with the induced penalty $\lambda/\varepsilon^{2}$ .

5% labels
Method	EC50			IC50
	Assay	Scaffold	Size	Assay	Scaffold	Size	Avg.
MVGRL	68.9 $\pm$ 3.7	52.1 $\pm$ 3.7	57.8 $\pm$ 2.9	67.1 $\pm$ 0.3	57.4 $\pm$ 1.7	62.1 $\pm$ 0.8	60.9
VGAE ( $\beta{=}0$ )	58.2 $\pm$ 2.8	50.0 $\pm$ 6.2	52.9 $\pm$ 4.1	61.5 $\pm$ 2.2	55.7 $\pm$ 2.2	56.8 $\pm$ 0.7	55.9
GraphCL	65.1 $\pm$ 2.0	54.6 $\pm$ 9.0	56.3 $\pm$ 1.8	67.0 $\pm$ 1.0	61.8 $\pm$ 1.9	61.5 $\pm$ 1.1	61.1
RGCL	69.5 $\pm$ 3.1	54.8 $\pm$ 7.5	56.5 $\pm$ 2.9	65.8 $\pm$ 0.1	60.0 $\pm$ 1.6	60.5 $\pm$ 1.1	61.2

Table 2: DrugOOD under limited labels. Test ROC-AUC (%), mean

\pm

std over three seeds (IC50-size VGAE at 20%: two), for linear probes on frozen representations. Bold : best per column; Avg.: mean over the six splits; shaded row: Scarce .

Appendix figures & tables10 assets

Supplementary material from the paper’s appendix.

Appendix

Dataset	Domain	$\alpha$	$\beta$	$K$	Test ROC-AUC (%)
EC50	Assay	0	10	3	69.84
		0	10	20	65.70
		0.5	10	20	65.63
		0	10	5	65.61
		0.5	10	3	64.28
		1	10	20	63.96

Appendix

Table 3: Top evaluated pseudo-environment configurations at 5% labels. Within each dataset–domain setting, the six displayed

(\alpha,\beta,K)

configurations are ordered from highest to lowest test ROC-AUC (%). Untested configurations are not interpolated.

Dataset	Domain	$\alpha$	$\beta$	$K$	Test ROC-AUC (%)
EC50	Assay	0	10	3	70.38
		0	10	10	68.45
		0	10	5	68.32
		0.5	10	5	67.17
		0	0.1	3	66.59
		0.5	1	20	66.39

Appendix

Table 4: Top evaluated pseudo-environment configurations at 10% labels. Within each dataset–domain setting, the six displayed

(\alpha,\beta,K)

configurations are ordered from highest to lowest test ROC-AUC (%). Untested configurations are not interpolated.

Dataset	Domain	$\alpha$	$\beta$	$K$	Test ROC-AUC (%)
EC50	Assay	0	10	3	73.07
		0.5	10	3	71.55
		0	10	5	69.67
		0.5	1	20	69.25
		0	1	3	68.31
		0.5	1	3	68.31

Appendix

Table 5: Top evaluated pseudo-environment configurations at 20% labels. Within each dataset–domain setting, the six displayed

(\alpha,\beta,K)

configurations are ordered from highest to lowest test ROC-AUC (%). Untested configurations are not interpolated.

$\alpha$	$\beta$	Test accuracy (%)
0	0.01	65.0
0	0.1	62.8
0	1	66.0
0	10	61.7
0.5	0.01	62.3
0.5	0.1	69.6

Appendix

Table 6: GOOD-Motif concept hyperparameter sensitivity. The evaluated configurations use fixed

K=8

and

k_{\rm adv}=2

. Test accuracy is the best-epoch linear-probe result.

Figure 5: Training dynamics on EC50-assay. The plot uses the validation-selected

(\alpha,\beta,K)=(0,10,3)

. Reconstruction, confusion, and adversary losses use solid, dashed, and dash-dotted lines with triangle, circle, and square markers, respectively. Lines are means over three seeds and shaded regions denote mean

\pm

sample standard deviation.

Figure 6: Training dynamics on EC50-scaffold. The validation-selected

(\alpha,\beta,K)

configuration is

(1,10,20)

. Lines and shaded regions follow the convention of Figure 5 .

Figure 7: Training dynamics on EC50-size. The validation-selected

(\alpha,\beta,K)

configuration is

(0.5,10,3)

. Lines and shaded regions follow the convention of Figure 5 .

Figure 8: Training dynamics on IC50-assay. The validation-selected

(\alpha,\beta,K)

configuration is

(1,10,20)

. Lines and shaded regions follow the convention of Figure 5 .

Figure 9: Training dynamics on IC50-scaffold. The validation-selected

(\alpha,\beta,K)

configuration is

(0.5,10,3)

. The vertical axis is truncated at 5 to show the stable-loss range; values above this limit are placed at the boundary and connected to the next sampled vertex. Lines and shaded regions follow the convention of Figure 5 .

Figure 10: Training dynamics on IC50-size. The validation-selected

(\alpha,\beta,K)

configuration is

(0.5,10,3)

. The vertical axis is logarithmic because reconstruction has a substantially larger transient scale. Lines and shaded regions follow the convention of Figure 5 .

OOD Generalization

Out-of-distribution generalization studies how machine-learning models perform when test data, domains, tasks, or environments differ from training.

See all 395 papers

Complementary Supervised and Self-Supervised Representations for Out-of-Distribution Graph Learning
Oct 6, 2026Qingying Hao, Zikang Chen, Chuxuan Hu +4Graph Neural Networks Graph Representation Learning
ShanghaiTech University Shanghai, China · University of Illinois Urbana-Champaign Urbana, Illinois, USA · The Pennsylvania State University University Park, Pennsylvania, USA
Adversarial Label Invariant Graph Data Augmentations for Out-of-Distribution Generalization
Apr 9, 2026Simon Zhang, Ryan P. DeMilt, Kun Jin +1OOD Generalization Covariate Shift
Department of Computer Science, Purdue University, West Lafayette, USA · Department of Computer Science and Engineering, The Ohio State University, Columbus Ohio, USA · Department of Industrial and Systems Engineering, The Ohio State University, Columbus Ohio, USA
SCL-GNN: Towards Generalizable Graph Neural Networks via Spurious Correlation Learning
Mar 9, 2026Yuxiang Zhang, Enyan DaiGraph Neural Networks Spurious Correlation Robustness
The Hong Kong University of Science and Technology (Guangzhou)

Oct 6, 2026J/K move · Enter open · S save

Complementary Supervised and Self-Supervised Representations for Out-of-Distribution Graph Learning

Qingying Hao, Zikang Chen, Chuxuan Hu, Jinyuan Jia+3

ShanghaiTech University Shanghai, China · University of Illinois Urbana-Champaign Urbana, Illinois, USA · The Pennsylvania State University University Park, Pennsylvania, USA

Graph Neural Networks Graph Representation Learning Graph SSL OOD Generalization

Open paper AskPDF arXiv

Out-of-distribution (OOD) generalization remains challenging for graph neural networks (GNNs), as graph distributions can vary substantially across time and domains. Supervised and self-supervised graph representation learning are guided by distinct objectives and offer different perspectives on graph representations. In this work, we study whether self-supervised representations (SSL) can provide complementary signals to improve supervised OOD node classification. We develop two backbone-agnostic frameworks that exploit such information at different stages of learning and prediction. Co-Train jointly learns supervised and SSL representations and adaptively integrates them during training, while Dual-Space Retrieval performs non-parametric prediction in the two representation spaces and combines their predictions through confidence-aware fusion at inference time. The supervised and SSL encoders are separately parameterized and need not share the same GNN architecture. We evaluate multiple GNN backbones and two distinct SSL objectives, DGI and GRACE, on four graph benchmarks spanning temporal and cross-domain distribution shifts. Extensive experiments show that Co-Train consistently outperforms strong supervised OOD baselines, while Dual-Space Retrieval achieves competitive performance as a flexible non-parametric alternative. Results across different backbones and SSL objectives, together with representation analyses and ablations, demonstrate that SSL representations provide complementary information to supervised representations and can improve OOD node classification across diverse settings.

Abstract

OOD Generalization

Explore similar work

Complementary Supervised and Self-Supervised Representations for Out-of-Distribution Graph Learning

Adversarial Label Invariant Graph Data Augmentations for Out-of-Distribution Generalization

SCL-GNN: Towards Generalizable Graph Neural Networks via Spurious Correlation Learning