v277: Proceedings of ML-DE 2025

2026-06-04 · Source: Proceedings of Machine Learning Research · Field: Science & Research — Artificial Intelligence & Machine Learning, Mathematics & Computational Sciences, Engineering & Applied Sciences · Depth: Expert, quick

Summary

Volume 277 compiles proceedings from the 2nd ECAI Workshop on "Machine Learning Meets Differential Equations: From Theory to Applications," held on October 26, 2025, at the University of Bologna. This collection features seven papers exploring advanced techniques at the intersection of these fields. Topics include learning non-Markovian dynamical systems using signature-based encoders, and TaylorNet, a method for learning Partial Differential Equations (PDEs) from non-grid data. Other contributions present Physics-Informed Graph Neural Networks for air pollution forecasting in the Netherlands, a unified framework for neural computation over time, and ExtremONet, an extreme-learning-based neural operator for identifying dynamical systems. The volume also covers Conditional Flow Matching for speech enhancement and Hamiltonian Normalizing Flows as kinetic PDE solvers, specifically applied to 1D Vlasov-Poisson Equations.

Key takeaway

For AI and Research Scientists exploring advanced modeling, this workshop volume offers a valuable overview of current research at the intersection of machine learning and differential equations. You should review these proceedings to identify novel techniques for learning dynamical systems, solving PDEs from non-grid data, or applying physics-informed neural networks to complex forecasting challenges. Consider the diverse approaches, from signature-based encoders to Hamiltonian Normalizing Flows, to inform your next project's methodological choices.

Key insights

The workshop explores diverse applications of machine learning to differential equations across various scientific and engineering domains.

Topics

• Machine Learning
• Differential Equations
• Dynamical Systems
• Physics-Informed Neural Networks
• Neural Operators
• Speech Enhancement

Code references

• eliottprdlx/Signature-Encoders-For-Dynamics-Learning

Best for: AI Scientist, Research Scientist

Related on AIssential

• Learning Physical Operators using Neural Operators — A physics-informed framework improves neural operator generalization and temporal continuity for PDE solving via operator splitting.
• #AAAI2026 invited talk: machine learning for particle physics — Machine learning, especially graph neural networks, can overcome limiting assumptions in particle tracking to discover new physics.
• Engineering Hybrid Physics-Informed Neural Networks for Next-Generation Electricity Systems: A State-of-the-Art Review — Physics-informed machine learning integrates domain physics into models, enhancing accuracy and efficiency for data-scarce electricity systems.
• A Machine Learning Approach to the Nirenberg Problem — A physics-informed neural network can effectively solve the Nirenberg problem and distinguish between realisable and non-realisable curvatures.
• Co-Learning Port-Hamiltonian Systems and Optimal Energy-Shaping Control — Co-learning pH system models and optimal EB-PBC policies via alternating optimization ensures stability and robustness in control.
• FLAME 2023: Diving into the Future of Fluid Dynamics and Machine Learning — AI techniques are transforming simulation by integrating physics-based modeling with data-driven machine learning.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Proceedings of Machine Learning Research.