Abstract: This report synthesises findings from 4 peer-reviewed papers addressing the following research question: Do lightweight graph contrastive learning methods (e.g., those using single-view augmentations) achieve comparable adversarial robustness to complex multi-view augmentation pipelines when evaluated.…

Abstract: This report synthesises findings from 14 peer-reviewed papers addressing the following research question: Can the integration of multimodal contrastive learning (e.g., combining text and graph data) improve the adversarial robustness of graph-based recommendation systems, as measured by AUC or. In the last few years,…

Abstract: This report synthesises findings from 15 peer-reviewed papers addressing the following research question: How does the adversarial robustness of graph contrastive learning models compare to traditional graph neural networks (GNNs) when evaluated on benchmark datasets like OGB, Cora, or Citeseer under. Deep models…

Abstract: This report synthesises findings from 10 peer-reviewed papers addressing the following research question: To what extent does the noise injection strategy in XSimGCL improve robustness against adversarial perturbations in user-item interaction graphs relative to traditional augmentation-based contrastive. Abstract…

Abstract: This report synthesises findings from 14 peer-reviewed papers addressing the following research question: Does the simplified noise injection approach in graph contrastive learning maintain ranking accuracy when scaled to extreme sparsity levels compared to heavy augmentation techniques in. Deep convolutional neural…

Abstract: This report synthesises findings from 6 peer-reviewed papers addressing the following research question: How does the inference efficiency of LightGCL scale with dataset size compared to SimGCL and DCL when evaluated on perturbed HOI datasets with robustness metrics. In recent years, neural architecture-based…

Abstract: This report synthesises findings from 1 peer-reviewed paper addressing the following research question: How does the robustness of contrastive learning frameworks like LightGCL, SimGCL, and DCL compare when evaluated on corrupted human-object interaction datasets using mAP@k metrics. Contrastive learning-based…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: What is the impact of adversarial training on the inference efficiency and throughput of large language models when deployed on code generation tasks. This article presents a comprehensive and practical guide for…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: What is the relative performance of LightGCL versus SimGCL in terms of recommendation accuracy (e.g., Recall@K, NDCG@K) when trained on large-scale sparse interaction graphs with varying levels of. Graph neural…

Abstract: This report synthesises findings from 2 peer-reviewed papers addressing the following research question: What is the impact of varying levels of occlusion noise on the performance of SimGCL and DCL when benchmarked against LightGCL using recall@k and mAP@k metrics on HOI detection datasets. Clustering of web…

Abstract: This report synthesises findings from 4 peer-reviewed papers addressing the following research question: What is the impact of domain adaptation techniques on the generalization performance of XSimGCL when applied to cross-domain datasets (e.g., from Reddit to DBLP) while preserving NDCG@10 scores above. The…

Abstract: This report synthesises findings from 8 peer-reviewed papers addressing the following research question: How robust is LightGCL's performance under adversarial graph perturbations (e.g., edge attacks) compared to other contrastive learning methods (e.g., GraCL, MVGRL) when evaluated using Hit Ratio@5. Abstract Data…

Abstract: This report synthesises findings from 10 peer-reviewed papers addressing the following research question: How does the contrastive learning performance of LightGCL compare to other GNN-based recommendation models (e.g., SGL, GCA) when evaluated on standard benchmarks like MovieLens-100K and Amazon Book. Graph neural…

Abstract: This report synthesises findings from 9 peer-reviewed papers addressing the following research question: How does the adversarial robustness of graph contrastive learning methods compare to vision-language contrastive models when evaluated on multimodal reasoning benchmarks under similar perturbation. In the last few…

Abstract: This report synthesises findings from 15 peer-reviewed papers addressing the following research question: Does graph sparsification in contrastive learning models maintain robustness against noise while improving throughput on sparse user-item graphs. Multilayer neural networks trained with the back-propagation…

Abstract: This report synthesises findings from 15 peer-reviewed papers addressing the following research question: How does the scaling of graph diffusion models compare to traditional graph neural networks (GNNs) in terms of inference efficiency and node classification accuracy when applied to large-scale graphs. Machine…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: What is the trade-off between clustering accuracy and computational efficiency when applying graph sparsification to contrastive learning-based recommenders. Abstract This paper critically examines model…

Abstract: This report synthesises findings from 13 peer-reviewed papers addressing the following research question: How does edge sparsification affect the inference latency and FLOPs of graph contrastive learning models for recommendation across datasets with varying density. Deep convolutional neural networks have performed…

Abstract: This report synthesises findings from 4 peer-reviewed papers addressing the following research question: What is the impact of different negative sampling strategies in contrastive graph learning on node clustering accuracy (NMI) and model convergence speed when applied to sparse versus dense regions of. Deformable…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: How does the integration of graph diffusion models with large language models (LLMs) impact the performance of template-based graph clustering when evaluated on node classification accuracy and. Abstract Large…

Abstract: This report synthesises findings from 8 peer-reviewed papers addressing the following research question: Which multimodal graph learning approaches achieve the highest node clustering accuracy (NMI) when integrating structural, textual, and numerical attributes in heterogeneous graphs like PDNS-Net, and. Feature…

Abstract: This report synthesises findings from 8 peer-reviewed papers addressing the following research question: To what extent do contrastive graph anomaly detection methods generalize across domains with varying attribute dimensionalities compared to supervised and semi-supervised benchmarks. Graphs naturally appear in…

Abstract: This report synthesises findings from 11 peer-reviewed papers addressing the following research question: Does integrating diffusion-based adversarial examples during training improve the pass@1 scores of multimodal code generation models on cross-domain reasoning tasks involving graph-structured inputs. This paper…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: What is the impact of contrastive pre-training on the inference efficiency and scalability of graph neural network anomaly detectors compared to reconstruction-based methods on large-scale attributed. In…

Abstract: This report synthesises findings from 3 peer-reviewed papers addressing the following research question: How does the performance of contrastive learning augmentation strategies in graph neural networks scale with increasing graph size and heterogeneity, as measured by node clustering accuracy (NMI) and. We trained a…