Researchers have introduced DyGFM, a novel Dynamic Graph Foundation Model designed to handle data from multiple domains. This model addresses the challenge of inconsistent semantic and temporal patterns across different domains, which typically hinders unified modeling and leads to negative knowledge transfer. DyGFM employs a dual-branch pre-training strategy for semantic-temporal decoupling and a cross-domain routing mechanism for divergence-aware expert selection to mitigate negative transfer during adaptation. The model also features a divergence-conditioned prompt generator for efficient downstream fine-tuning, showing superior performance on node classification and link prediction tasks compared to existing methods. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Introduces a new foundation model architecture for dynamic graphs, potentially improving performance and generalization across diverse real-world datasets.
RANK_REASON The cluster contains an academic paper detailing a new model architecture and its performance on benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]