Papers
Topics
Authors
Recent
Search
2000 character limit reached

Learning dissipative Hamiltonian dynamics with reproducing kernel Hilbert spaces and random Fourier features

Published 24 Oct 2024 in cs.LG and cs.RO | (2410.18656v1)

Abstract: This paper presents a new method for learning dissipative Hamiltonian dynamics from a limited and noisy dataset. The method uses the Helmholtz decomposition to learn a vector field as the sum of a symplectic and a dissipative vector field. The two vector fields are learned using two reproducing kernel Hilbert spaces, defined by a symplectic and a curl-free kernel, where the kernels are specialized to enforce odd symmetry. Random Fourier features are used to approximate the kernels to reduce the dimension of the optimization problem. The performance of the method is validated in simulations for two dissipative Hamiltonian systems, and it is shown that the method improves predictive accuracy significantly compared to a method where a Gaussian separable kernel is used.

Authors (2)
Definition Search Book Streamline Icon: https://streamlinehq.com
References (28)
  1. Control-Oriented Learning of Lagrangian and Hamiltonian Systems. In American Control Conference (ACC), 520–525. 10.23919/ACC.2018.8431726.
  2. Arnold, V.I. (1989). Mathematical Methods of Classical Mechanics. Graduate Texts in Mathematics. Springer, New York, 2 edition. 10.1007/978-1-4757-2063-1.
  3. Aronszajn, N. (1950). Theory of Reproducing Kernels. Transactions of the American Mathematical Society, 68(3), 337–404.
  4. Nonparametric adaptive control and prediction: theory and randomized algorithms. Journal of Machine Learning Research, 23(281), 1–46.
  5. Random Fourier Features For Operator-Valued Kernels. In Proceedings of The 8th Asian Conference on Machine Learning, volume 63 of Proceedings of Machine Learning Research, 110–125. PMLR.
  6. Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control. Cambridge University Press, 2 edition. 10.1017/9781009089517.
  7. Symplectic Recurrent Neural Networks. In International Conference on Learning Representations.
  8. Fuselier, E.J. (2006). Refined Error Estimates for Matrix-Valued Radial Basis Functions. PhD thesis, Texas A&M University.
  9. Helmholtz decomposition and potential functions for n𝑛nitalic_n-dimensional analytic vector fields. Journal of Mathematical Analysis and Applications, 525(2), 127138. 10.1016/j.jmaa.2023.127138.
  10. Classical Mechanics. Pearson, 3 edition.
  11. Hamiltonian Neural Networks. In Proceedings of the Conference on Neural Information Processing Systems (NeurIPS).
  12. Geometric Numerical Integration: Structure-Preserving Algorithms for Ordinary Differential Equations. Springer Series in Computational Mathematics. Springer Berlin, Heidelberg, 2 edition. 10.1007/3-540-30666-8.
  13. Kohavi, R. (1995). A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection. In Proceedings of the 14th International Joint Conference on Artificial Intelligence, IJCAI-95, 1137–1143.
  14. Reproducing Kernel Hilbert Spaces With Odd Kernels in Price Prediction. IEEE Transactions on Neural Networks and Learning Systems, 23(10), 1564–1573. 10.1109/TNNLS.2012.2207739.
  15. Lowitzsch, S. (2002). Approximation and Interpolation Employing Divergence-free Radial Basis Functions with Applications. Ph.D. thesis, Texas A&M University.
  16. On Learning Vector-Valued Functions. Neural Computation, 17(1), 177–204. 10.1162/0899766052530802.
  17. Minh, H.Q. (2016). Operator-Valued Bochner Theorem, Fourier Feature Maps for Operator-Valued Kernels, and Vector-Valued Learning. 10.48550/arXiv.1608.05639. ArXiv preprint arXiv:1608.05639 [cs.LG].
  18. Vector-valued Manifold Regularization. In Proceedings of the 28th International Conference on Machine Learning (ICML1 1), 57–64. Omnipress. 10.5555/3298023.3298180.
  19. Control by interconnection and standard passivity-based control of port-Hamiltonian systems. IEEE Transactions on Automatic Control, 53(11), 2527–2542.
  20. Uniform Approximation of Functions with Random Bases. In 2008 46th Annual Allerton Conference on Communication, Control, and Computing, 555–561. 10.1109/ALLERTON.2008.4797607.
  21. A Generalized Representer Theorem. In Computational Learning Theory, 416–426. Springer Berlin Heidelberg. 10.1007/3-540-44581-1_27.
  22. Learning Contracting Vector Fields For Stable Imitation Learning. 10.48550/arXiv.1804.04878. ArXiv preprint arXiv:1804.04878 [cs.RO].
  23. Learning stabilizable nonlinear dynamics with contraction-based regularization. The International Journal of Robotics Research, 40(10-11), 1123–1150. 10.1177/0278364920949931.
  24. Learning of Hamiltonian Dynamics with Reproducing Kernel Hilbert Spaces. 10.48550/arXiv.2312.09734. ArXiv preprint arXiv:2312.09734 [cs.RO].
  25. Learning Hamiltonian Dynamics with Reproducing Kernel Hilbert Spaces and Random Features. 10.48550/arXiv.2404.07703. ArXiv preprint arXiv:2404.07703 [cs.LG].
  26. Dissipative Hamiltonian Neural Networks: Learning Dissipative and Conservative Dynamics Separately. 10.48550/arXiv.2201.10085. ArXiv preprint arXiv:2201.10085 [cs.LG].
  27. Dissipative SymODEN: Encoding Hamiltonian Dynamics with Dissipation and Control into Deep Learning. 10.48550/arXiv.2002.08860. ArXiv preprint arXiv:2002.08860 [cs.LG].
  28. Symplectic ODE-Net: Learning Hamiltonian Dynamics with Control. In Proceedings of the International Conference on Learning Representations (ICLR).

Summary

No one has generated a summary of this paper yet.

Sign Up to Summarize

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Generate Now

Continue Learning

We haven't generated follow-up questions for this paper yet.

Generate Now

Collections

Sign up for free to add this paper to one or more collections.

Sign Up