Hands-on Machine Learning in Fluid Dynamics 2027
DATE: 08 - 12 March 2027
LOCATION: online / on-site - Waterloosesteenweg, 72, 1640 Sint-Genesius-Rode, Belgium - near Brussels
ABOUT THE LECTURE SERIES

Now in its seventh edition, Hands-on Machine Learning for Fluid Dynamics is a one-week intensive course given by the ML4F — Machine Learning for Fluid Systems group. Previous editions have welcomed more than 600 participants from academia, research institutes and industry.
The course is taught by current researchers and alumni of ML4F and draws on their research in fluid and thermo-fluid systems. Lectures connect the methods to practical problems, while hands-on Python sessions give participants the opportunity to implement and apply them.
The course explores how machine learning can help us understand, predict and control fluid flows. Participants learn to build models from data, quantify uncertainty and incorporate physical knowledge using regression, kernel methods and neural networks. Optimization, dimensionality reduction and reduced-order modeling provide tools for working with complex flow data, while turbulence modeling, system identification and data assimilation connect these methods to fluid-dynamics problems. Digital twinning, reinforcement learning and reinforcement twinning bring modeling and control together, with a final session examining the potential of AI agents to design flow controllers.
The course welcomes graduate students, researchers and professionals with a range of backgrounds. It builds from core concepts to advanced methods, combining theoretical foundations with practical examples from fluid dynamics. Exercises and tutorials allow participants to explore the material at their own pace and continue learning after the course.
Course directors
The course was designed by Prof. Miguel A. Mendez (Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid), who has directed it since 2020. For the 2027 edition, he is joined as co-director by Dr. Romain Poletti (UC Davis and von Karman Institute).
PROGRAMME
Monday, 8 March 2027: Getting Started with Machine Learning and Optimization
08:30 – 09:00 Registration and Welcome Coffee
09:00 – 09:15 Welcome Address and Course Overview
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid; Romain Poletti, UC Davis and von Karman Institute
09:15 – 10:45 Lecture 1: Machine Learning for Fluid Dynamics — Concepts and Applications
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
10:45 – 11:15 Coffee Break
11:15 – 12:45 Lecture 2: Fitting Data and Quantifying Uncertainty
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
12:45 – 14:00 Lunch Break
14:00 – 15:30 Lecture 3: Gradient-Based Optimization — From Principles to Practice
Romain Poletti, UC Davis and von Karman Institute
15:30 – 16:00 Coffee Break
16:00 – 17:30 Lecture 4: Bio-Inspired Optimization — From Genetic Algorithms to Particle Swarms
Romain Poletti, UC Davis and von Karman Institute
17:30 Reception
Tuesday, 9 March 2027: Learning from Data with Basis Functions, Kernels and Neural Networks
09:00 – 10:30 Lecture 5: Radial Basis Functions — Bringing Physics into Regression
Manuel Ratz, Université libre de Bruxelles and von Karman Institute
10:30 – 11:00 Coffee Break
11:00 – 12:30 Lecture 6: Bayesian Inference and Kernel-Based Learning
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
12:30 – 14:00 Lunch Break
14:00 – 15:30 Lecture 7: Gaussian Processes for Multi-Fidelity Learning and Bayesian Optimization
Yannick Lecomte, Université libre de Bruxelles and von Karman Institute
15:30 – 16:00 Coffee Break
16:00 – 17:30 Lecture 8: Deep Learning and Physics-Informed Neural Networks
Jan Van Den Berghe, von Karman Institute
17:30 End of Day
Wednesday, 10 March 2027: Learning and Predicting Fluid Dynamics from Data
09:00 – 10:30 Lecture 9: Learning Turbulence Closures from Data
Matilde Fiore, von Karman Institute
10:30 – 11:00 Coffee Break
11:00 – 12:30 Lecture 10: The Adjoint Method — Gradients for Dynamical Systems
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
12:30 – 14:00 Lunch Break
14:00 – 15:30 Lecture 11: Learning Dynamical Systems and Forecasting Their Evolution
Samuel Ahizi, Universidad Carlos III de Madrid and von Karman Institute
15:30 – 16:00 Coffee Break
16:00 – 17:30 Lecture 12: Kalman Filters — Combining Models and Measurements
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
17:30 End of Day
Thursday, 11 March 2027: Finding Low-Dimensional Structure in Flow Data
09:00 – 10:30 Lecture 13: From POD to Linear Autoencoders — Finding Patterns in Flow Data
Romain Poletti, UC Davis and von Karman Institute
10:30 – 11:00 Coffee Break
11:00 – 12:30 Lecture 14: Beyond Linear Models — Manifold Learning and Dimensionality Reduction
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
12:30 – 14:00 Lunch Break
14:00 – 15:30 Lecture 15: Variational Autoencoders — Learning Probabilistic Representations
Joachim Dominique, Cenaero
15:30 – 16:00 Coffee Break
16:00 – 17:30 Lecture 16: From Spectral Methods to Data-Driven Reduced-Order Models
Romain Poletti, UC Davis and von Karman Institute
Friday, 12 March 2027: Learning to Control Flows — Reinforcement Learning and Digital Twins
09:00 – 10:30 Lecture 17: From Optimal Control to Reinforcement Learning
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
10:30 – 11:00 Coffee Break
11:00 – 12:30 Lecture 18: Actor–Critic Methods — Learning Control Policies
Romain Poletti, UC Davis and von Karman Institute
12:30 – 14:00 Lunch Break
14:00 – 15:30 Lecture 19: Reinforcement Twinning — From Digital Twins to Model-Based Control
Miguel A. Mendez, Université libre de Bruxelles, von Karman Institute and Universidad Carlos III de Madrid
15:30 – 16:00 Coffee Break
16:00 – 17:30 Lecture 20: Can Frontier AI Agents Design Flow Controllers? An Open Benchmark
Lorenzo Schena, Vrije Universiteit Brussel and von Karman Institute
17:30 End of Course
FEE AND REGISTRATION
- Deadline for on-site registration: 22 February 2027
- Deadline for online registration: 1 March 2027
- Discount of 10% in case of online participation*
- Discount of 30% for citizens from NATO countries*
- Discount of 40% for citizens from NATO nations supporting the financing of VKI*
* applicable on the price listed below
Early Bird Registration until 08 January 2027
- Undergraduate students: €315
- PhD students: €864
- Staff from recognized universities / research center: €1368
- Staff from commercial organizations: €1728
Late Registration after 08 January 2027
- Undergraduate students: €350
- PhD students: €960
- Staff from recognized universities / research center: €1520
- Staff from commercial organizations: €1920
BOOK
Also available on Amazon the book "Machine Learning for Fluid Dynamics Paperback", English edition by M. A. Mendez (Editor), A. Parente (Editor), S. Brunton (Contributor), P. Cinnella (Contributor), L. Magri (Contributor), O. Semeraro (Contributor), S. Discetti (Contributor), S. Le Clainche (Contributor), P. Bekemeyer (Contributor) and O. San (Contributor). This book collects 12 lectures from the recent VKI lecture series "Machine Learning for Fluid Dynamics," jointly organized by VKI and ULB in February 2024.
The lecture series explored how modern machine-learning methods can be applied to study and solve challenges in fluid dynamics. In line with the VKI lecture series tradition, it brings together leading experts to cover both foundational concepts and state-of-the-art applications.
Starting with the fundamentals of regression and dimensionality reduction, the book progresses to advanced topics such as turbulence modeling, flow control, and digital twin development for a variety of multi-physics applications. It is a valuable resource for students, researchers, and professionals eager to leverage machine learning for enhanced simulations, predictions, and engineering solutions.
Product details
- Publisher : von Karman Institute for Fluid Dynamics
- Publication date : 21 Oct. 2025
- Language : English
- Print length : 486 pages
- ISBN-10 : 2875162098
- ISBN-13 : 978-2875162090
- Item weight : 943 g
- Dimensions : 16.99 x 2.79 x 24 cm



