Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AGMixup: Adaptive Graph Mixup for Semi-supervised Node Classification

Weigang Lu, Ziyu Guan, Wei Zhao, Yaming Yang, Yibing Zhan, Yiheng Lu, Dapeng Tao

TL;DR

An Adaptive Graph Mixup framework for semi-supervised node classification is proposed, which introduces a subgraph-centric approach, which treats each subgraph similarly to how images are handled in Euclidean domains, thus facilitating a more natural integration of mixup into graph-based learning.

Abstract

Mixup is a data augmentation technique that enhances model generalization by interpolating between data points using a mixing ratio $λ$ in the image domain. Recently, the concept of mixup has been adapted to the graph domain through node-centric interpolations. However, these approaches often fail to address the complexity of interconnected relationships, potentially damaging the graph's natural topology and undermining node interactions. Furthermore, current graph mixup methods employ a one-size-fits-all strategy with a randomly sampled $λ$ for all mixup pairs, ignoring the diverse needs of different pairs. This paper proposes an Adaptive Graph Mixup (AGMixup) framework for semi-supervised node classification. AGMixup introduces a subgraph-centric approach, which treats each subgraph similarly to how images are handled in Euclidean domains, thus facilitating a more natural integration of mixup into graph-based learning. We also propose an adaptive mechanism to tune the mixing ratio $λ$ for diverse mixup pairs, guided by the contextual similarity and uncertainty of the involved subgraphs. Extensive experiments across seven datasets on semi-supervised node classification benchmarks demonstrate AGMixup's superiority over state-of-the-art graph mixup methods. Source codes are available at \url{https://github.com/WeigangLu/AGMixup}.

AGMixup: Adaptive Graph Mixup for Semi-supervised Node Classification

TL;DR

An Adaptive Graph Mixup framework for semi-supervised node classification is proposed, which introduces a subgraph-centric approach, which treats each subgraph similarly to how images are handled in Euclidean domains, thus facilitating a more natural integration of mixup into graph-based learning.

Abstract

Mixup is a data augmentation technique that enhances model generalization by interpolating between data points using a mixing ratio in the image domain. Recently, the concept of mixup has been adapted to the graph domain through node-centric interpolations. However, these approaches often fail to address the complexity of interconnected relationships, potentially damaging the graph's natural topology and undermining node interactions. Furthermore, current graph mixup methods employ a one-size-fits-all strategy with a randomly sampled for all mixup pairs, ignoring the diverse needs of different pairs. This paper proposes an Adaptive Graph Mixup (AGMixup) framework for semi-supervised node classification. AGMixup introduces a subgraph-centric approach, which treats each subgraph similarly to how images are handled in Euclidean domains, thus facilitating a more natural integration of mixup into graph-based learning. We also propose an adaptive mechanism to tune the mixing ratio for diverse mixup pairs, guided by the contextual similarity and uncertainty of the involved subgraphs. Extensive experiments across seven datasets on semi-supervised node classification benchmarks demonstrate AGMixup's superiority over state-of-the-art graph mixup methods. Source codes are available at \url{https://github.com/WeigangLu/AGMixup}.

Paper Structure

This paper contains 50 sections, 7 equations, 15 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Mixup on Images.
  4. Mixup on Graphs.
  5. Comparison to SOTA Graph Mixup.
  6. Methodology
  7. Preliminaries
  8. Notations.
  9. GNN Revisit.
  10. Mixup Revisit.
  11. Proposed Method: AGMixup
  12. Subgraph-Centric Mixup.
  13. Contextual Similarity-aware $\lambda$ Initialization.
  14. Uncertainty-aware $\lambda$ Adjustment.
  15. Training with AGMixup.
  16. ...and 35 more sections

Figures (15)

  • Figure 1: Seamlessly integrating mixup from image domain into graph domain with our AGMixup.
  • Figure 2: Difference between AGMixup and SOTA graph mixup methods.
  • Figure 3: Prediction errors in-between training data. Our AGMixup significantly reduces the rate of prediction misses, indicating superior interpolation capability.
  • Figure 4: Confidence comparison. Our AGMixup makes the model more "confident" in the predictions by paying more attention to those underrepresented samples.
  • Figure 5: Efficacy analysis using GCN as backbone model.
  • ...and 10 more figures