RA-GCN: Graph convolutional network for disease prediction problems with imbalanced data

Ghorbani, Mahsa; Kazi, Anees; Baghshah, Mahdieh Soleymani; Rabiee, Hamid R.; Navab, Nassir

doi:10.1016/j.media.2021.102272

Cited by 61 publications

(13 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The fraction of the majority classes is µ, and for all experiments, we set µ = 0.5 and round down the result µ * k. The training loss L Q from section 2.4 is adopted. We implement two groups of baselines for comparison: (1) Popular quantity-imbalance methods for general scenarios: Re-weight [20] (RW), Focal Loss [28] (Focal) and Class Balanced Loss [9] (CB); (2) Graph-specific quantity-imbalance methods: DR-GCN [44], RA-GCN [14] and GraphSMOTE [58]. To jointly handle the topology-and quantity-imbalance issues and demonstrate the orthogonality of them, we combine our ReNode method with these three general quantity-imbalance methods (RW, Focal, CB) 4 .…”

Section: Renode For the Compound Scene Of Tinl And Qinlmentioning

confidence: 99%

“…DR-GCN [44] propose two types of regularization to tackle quantity imbalance: class-conditioned adversarial training and unlabeled nodes latent distribution constraint. RA-GCN [14] propose to automatically learn to weight the training samples in different classes in an adversarial training manner. AdaGCN Shi et al [45] propose to leverage the boosting algorithm to handle the quantity-imbalance issue for the node classification task.…”

Section: Related Workmentioning

confidence: 99%

“…Similar to other data structures, the representation learning for node classification faces the challenge of quantity-imbalance issue, where the labeling size varies among classes and the decision boundaries of trained classifiers are mainly decided by the majority classes [55]. There have been a series of studies [44,14,58] handling the Quantity-Imbalance Node Representation Learning (short as QINL). However, different with other data structures, graph-structured data suffers from another aspect of the imbalance problem: the imbalance caused by the asymmetric and uneven topology of labeled nodes, where the decision boundaries are driven by the labeled nodes close to the topological class boundaries (left of Figure 1) thus interfering with the model learning.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Topology-Imbalance Learning for Semi-Supervised Node Classification

Chen,

Lin,

Zhao

et al. 2021

Preprint

View full text Add to dashboard Cite

The class imbalance problem, as an important issue in learning node representations, has drawn increasing attention from the community. Although the imbalance considered by existing studies roots from the unequal quantity of labeled examples in different classes (quantity imbalance), we argue that graph data expose a unique source of imbalance from the asymmetric topological properties of the labeled nodes, i.e., labeled nodes are not equal in terms of their structural role in the graph (topology imbalance). In this work, we first probe the previously unknown topology-imbalance issue, including its characteristics, causes, and threats to semisupervised node classification learning. We then provide a unified view to jointly analyzing the quantity-and topology-imbalance issues by considering the node influence shift phenomenon with the Label Propagation algorithm. In light of our analysis, we devise an influence conflict detection-based metric Totoro to measure the degree of graph topology imbalance and propose a model-agnostic method ReNode to address the topology-imbalance issue by re-weighting the influence of labeled nodes adaptively based on their relative positions to class boundaries. Systematic experiments demonstrate the effectiveness and generalizability of our method in relieving topology-imbalance issue and promoting semi-supervised node classification. The further analysis unveils varied sensitivity of different graph neural networks (GNNs) to topology imbalance, which may serve as a new perspective in evaluating GNN architectures. 1

show abstract

Section: Renode For the Compound Scene Of Tinl And Qinlmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Topology-Imbalance Learning for Semi-Supervised Node Classification

Chen,

Lin,

Zhao

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Then, the pseudo labels are used to train a student graph for Autism spectrum disorder or Alzheimer's disease prediction. RA‐GCN [28], on the other hand, addressed the imbalanced class distribution in the medical data by representing each class by a graph‐based neural network responsible for the weighting of class samples. The whole architecture, then, is trained in an adversarial manner such that the classifier adapts itself with the attention to rare cases.…”

Section: Introductionmentioning

confidence: 99%

ROAM: Random layer mixup for semi‐supervised learning in medical images

Bdair

Wiestler

Navab

et al. 2022

IET Image Processing

Self Cite

View full text Add to dashboard Cite

Medical image segmentation is one of the major challenges addressed by machine learning methods. However, these methods profoundly depend on a large amount of annotated data, which is expensive and time-consuming. Semi-supervised learning (SSL) approaches this by leveraging an abundant amount of unlabeled data. Recently, MixUp regularizer has been introduced to SSL methods by augmenting the model with new data points through linear interpolation at the input space. While this provides the model with new data, it is limited and may lead to inconsistent soft labels. It is argued that the linear interpolation at different representations provides the network with novel training signals and overcomes the inconsistency of the soft labels. This paper proposes ROAM as an SSL method that explores the manifold and performs linear interpolation on randomly selected layers to generate virtual data that has never been seen before, which encourages the network to be less confident for interpolated points. Hence it avoids overfitting, enhances the generalization, and shows less sensitivity to the domain shift. Extensive experiments are conducted on publicl datasets on whole-brain and lung segmentation. ROAM achieves state-of-theart results in fully supervised (89.5%) and semi-supervised (87.0%) settings with relative improvements up to 2.40% and 16.50%, respectively.

show abstract

“…Indeed, GCNs are widely used for disease diagnosis [5], [6], due to the biomarkers across all subjects are critical for recognizing the common patterns associated with diseases. Previous studies have explored the development of GCNs for various prediction tasks on fMRI data [3], [5], [7], [8]. For example, Parisot et al [5] apply vanilla GCN for supervised disease prediction with fMRI data.…”

Section: Introductionmentioning

confidence: 99%

GATE: Graph CCA for Temporal SElf-supervised Learning for Label-efficient fMRI Analysis

Peng¹,

Wang²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

In this work, we focus on the challenging task, neuro-disease classification, using functional magnetic resonance imaging (fMRI). In population graph-based disease analysis, graph convolutional neural networks (GCNs) have achieved remarkable success. However, these achievements are inseparable from abundant labeled data and sensitive to spurious signals. To improve fMRI representation learning and classification under a label-efficient setting, we propose a novel and theory-driven self-supervised learning (SSL) framework on GCNs, namely Graph CCA for Temporal sElf-supervised learning on fMRI analysis (GATE). Concretely, it is demanding to design a suitable and effective SSL strategy to extract formation and robust features for fMRI. To this end, we investigate several new graph augmentation strategies from fMRI dynamic functional connectives (FC) for SSL training. Further, we leverage canonicalcorrelation analysis (CCA) on different temporal embeddings and present the theoretical implications. Consequently, this yields a novel two-step GCN learning procedure comprised of (i) SSL on an unlabeled fMRI population graph and (ii) fine-tuning on a small labeled fMRI dataset for a classification task. Our method is tested on two independent fMRI datasets, demonstrating superior performance on autism and dementia diagnosis.

show abstract

RA-GCN: Graph convolutional network for disease prediction problems with imbalanced data

Cited by 61 publications

References 27 publications

Topology-Imbalance Learning for Semi-Supervised Node Classification

Topology-Imbalance Learning for Semi-Supervised Node Classification

ROAM: Random layer mixup for semi‐supervised learning in medical images

GATE: Graph CCA for Temporal SElf-supervised Learning for Label-efficient fMRI Analysis

Contact Info

Product

Resources

About