New GKAN-ODE model excels at discovering graph dynamical system equations

Researchers have developed a new evaluation pipeline to rigorously assess symbolic regression models for discovering governing equations in graph dynamical systems. This framework moves beyond simple fitting metrics to evaluate discovered laws based on their long-term trajectory stability and generalization to unseen graph topologies. A novel model, the Graph Kolmogorov-Arnold Network-ODE (GKAN-ODE), was introduced and demonstrated superior performance, recovering exact ground-truth equations and achieving significantly lower trajectory errors compared to baseline methods on both synthetic and real-world datasets. AI

IMPACT This research could lead to more accurate and generalizable AI models for scientific discovery in complex systems.

RANK_REASON The cluster describes a new academic paper introducing a novel model and evaluation framework for a specific scientific problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• DagsHub
• Graph Kolmogorov-Arnold Network-ODE
• Hugging Face
• Kolmogorov--Arnold Networks
• MLP-based architectures
• Riccardo Cappi

• paper
• model release
• other

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