New method enhances explainability for Temporal Graph Neural Networks

Researchers have developed a new method to explain the workings of Event-based Temporal Graph Neural Networks (ETGNNs). Current methods only analyze a portion of the information flow, missing crucial pathways through event-induced variables that capture long-range temporal dependencies. The proposed approach, built on the Normalized Relevance Measure (NRM) framework, analyzes the entire information flow, including event embeddings and event-induced variables, to provide more comprehensive and interpretable explanations. This method has been evaluated on synthetic and real-world datasets, demonstrating superior performance over existing techniques. AI

IMPACT Enhances interpretability of complex temporal graph neural networks, aiding in debugging and trust for applications like social network analysis and epidemic tracing.

RANK_REASON The cluster contains a research paper detailing a new method for explaining AI models.

Read on arXiv cs.LG →

• alphaXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• epidemic tracing
• ETGNNs
• Event-based Temporal Graph Neural Networks
• Gotit.pub
• Hugging Face
• Litmaps
• Normalized Relevance Measure
• political event forecasting
• Recommender Systems
• ScienceCast
• scite Smart Citations
• social network analysis

• paper
• other

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →