New framework enhances physics-informed neural networks with information propagation paths

Researchers have developed a new framework for physics-informed neural networks (PINNs) that addresses limitations in how these networks handle partial differential equations. The proposed multi-dimensional training-priority weighting system accounts for the physical propagation of information, prioritizing learning from source regions to dependent regions across spatial, temporal, and boundary dimensions. This approach, analyzed using neural tangent kernel dynamics, aims to improve training stability and prediction accuracy by aligning the network's learning process with physical principles, showing consistent improvements on benchmark cases without altering the network architecture. AI

IMPACT Improves the accuracy and stability of neural networks used for solving complex physical simulations.

RANK_REASON Academic paper detailing a new framework for physics-informed neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• Neural tangent kernel
• partial differential equations
• physics-informed neural networks
• ScienceCast
• scite Smart Citations

• paper
• infra

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