New framework DGOTTA enhances GNN adaptation on dynamic graphs

Researchers have introduced DGOTTA, a novel framework designed for temporal memory-aware online test-time adaptation on dynamic graphs. This approach addresses the challenge of adapting graph neural networks (GNNs) to evolving graph structures and node semantics during testing, a problem not adequately covered by existing methods for static graphs. DGOTTA incorporates modules for temporal-aware augmentation, memory-aware prediction to prevent catastrophic forgetting, and consistency-guided online adaptation to ensure temporal alignment and memory smoothness. Experiments across various datasets and GNN architectures demonstrate DGOTTA's effectiveness in improving generalization under distribution shifts. AI

IMPACT This research could improve the robustness and generalization of AI models used in dynamic environments, such as real-time fraud detection or social network analysis.

RANK_REASON The item is an academic paper detailing a new method for adapting graph neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• DGOTTA
• Dynamic graph neural network for fake news detection
• Gotit.pub
• graph neural network
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →