Towards Data Science is a community-powered publication that showcases work in data science, machine learning and artificial intelligence. Every day newcomers, seasoned researchers and industry practitioners publish tutorials, research notes and real-world case studies that help the field move forward.

Towards Data Science

Physics-Informed Neural Networks (PINNs) are often overparameterized by orders of magnitude. Through experiments on phase field fracture, Burgers' equation, and hyperelastic deformation problems, networks with 50-400× fewer parameters achieved comparable accuracy to those in published literature. Small networks suffice for low-frequency solution fields with localized features like cracks and shocks, reducing computational cost significantly. However, high-frequency oscillatory functions still require larger networks. The key insight: use the smallest network that maintains accuracy, as training cost scales with parameter count.

On the Possibility of Small Networks for Physics-Informed Learning