Timezone: »

 
Poster
Approximately Equivariant Networks for Imperfectly Symmetric Dynamics
Rui Wang · Robin Walters · Rose Yu

Tue Jul 19 03:30 PM -- 05:30 PM (PDT) @ Hall E #413

Incorporating symmetry as an inductive bias into neural network architecture has led to improvements in generalization, data efficiency, and physical consistency in dynamics modeling. Methods such as CNN or equivariant neural networksuse weight tying to enforce symmetries such as shift invariance or rotational equivariance. However, despite the fact that physical laws obey many symmetries, real-world dynamical data rarely conforms to strict mathematical symmetry either due to noisy or incomplete data or to symmetry breaking features in the underlying dynamical system. We explore approximately equivariant networks which are biased towards preserving symmetry but are not strictly constrained to do so. By relaxing equivariance constraints, we find that our models can outperform both baselines with no symmetry bias and baselines with overly strict symmetry in both simulated turbulence domains and real-world multi-stream jet flow.

Author Information

Rui Wang (University of California, San Diego)
Robin Walters (Northeastern University)
Rose Yu (University of California, San Diego)

Dr. Rose Yu is an assistant professor at the UC San Diego, Department of Computer Science and Engineering. Her research focuses on advancing machine learning techniques for large-scale spatiotemporal data analysis, with applications to sustainability, health, and physical sciences. A particular emphasis of her research is on physics-guided AI which aims to integrate first principles with data-driven models. Among her awards, she has won Faculty Research Award from Facebook, Google, Amazon, and Adobe, several Best Paper Awards, and the Best Dissertation Award in USC.

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors