MEGA Hub

Dynamic Graph Prompting via Topology-Routed Mixed-Curvature Experts

Authors

Do you know Quanxin Wang?You can claim authorship or link another user.Do you know Xuanting Xie?You can claim authorship or link another user.Do you know Bingheng Li?You can claim authorship or link another user.Do you know Xingtong Yu?You can claim authorship or link another user.Do you know Shuo Wang?You can claim authorship or link another user.Do you know Ruiyi Fang?You can claim authorship or link another user.Do you know Zhao Kang?You can claim authorship or link another user.

Abstract

Dynamic graph prompting freezes a pre-trained temporal backbone and adapts it to label-scarce downstream tasks using lightweight prompts. However, existing methods operate within a single, fixed embedding space. In this work, we reveal that temporal shifts in local clustering and degree heterogeneity actively reorganize the edge curvature spectrum---indicating that the optimal representation geometry dynamically evolves with local topology over time. We formalize this unaddressed mismatch as geometry under-adaptation. To overcome this limitation, we propose CurvPrompt, a topology-routed geometry prompting framework for dynamic graphs. Instead of relying on a single space, CurvPrompt maintains a bank of curvature-diverse Riemannian experts, each paired with a learnable prompt. A topology-aware gate dynamically routes each node--time instance to a sparse subset of experts, constructing a personalized mixed-curvature representation. To ensure parameter efficiency and training stability under extreme label scarcity, CurvPrompt employs soft routing during pre-training to build a continuous topology--geometry mapping, and transitions to hard Top-K routing with uniform weights during downstream adaptation. Extensive experiments across four benchmark datasets show that CurvPrompt significantly advances few-shot link prediction while delivering strong, consistent performance on node classification tasks, validating the necessity of geometry-adaptive prompting.

Community

00

Publication notes

Author note
Suggestions and comments are welcomed