Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
125 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
42 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Deep Kernel Learning of Dynamical Models from High-Dimensional Noisy Data (2208.12975v3)

Published 27 Aug 2022 in cs.LG and stat.ML

Abstract: This work proposes a Stochastic Variational Deep Kernel Learning method for the data-driven discovery of low-dimensional dynamical models from high-dimensional noisy data. The framework is composed of an encoder that compresses high-dimensional measurements into low-dimensional state variables, and a latent dynamical model for the state variables that predicts the system evolution over time. The training of the proposed model is carried out in an unsupervised manner, i.e., not relying on labeled data. Our learning method is evaluated on the motion of a pendulum -- a well studied baseline for nonlinear model identification and control with continuous states and control inputs -- measured via high-dimensional noisy RGB images. Results show that the method can effectively denoise measurements, learn compact state representations and latent dynamical models, as well as identify and quantify modeling uncertainties.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (50)
  1. Data-driven science and engineering: Machine learning, dynamical systems, and control (Cambridge University Press, 2022).
  2. Machine learning, vol. 1 (McGraw-hill New York, 1997).
  3. Deep learning (MIT press, 2016).
  4. Reinforcement learning: An introduction (MIT press, 2018).
  5. Deep reinforcement learning: A brief survey. \JournalTitleIEEE Signal Processing Magazine 34, 26–38 (2017).
  6. State representation learning for control: An overview. \JournalTitleNeural Networks 108, 379–392 (2018).
  7. Unsupervised representation learning in deep reinforcement learning: A review. \JournalTitlearXiv preprint arXiv:2208.14226 (2022).
  8. Reduced order methods for modeling and computational reduction, vol. 9 (Springer, 2014).
  9. Reduced basis methods for time-dependent problems. \JournalTitleActa Numerica 31, 265–345 (2022).
  10. Model predictive control (Springer science & business media, 2013).
  11. Singular value decomposition and principal component analysis. In A practical approach to microarray data analysis, 91–109 (Springer, 2003).
  12. Principal component analysis. \JournalTitleChemometrics and intelligent laboratory systems 2, 37–52 (1987).
  13. The proper orthogonal decomposition in the analysis of turbulent flows. \JournalTitleAnnual review of fluid mechanics 25, 539–575 (1993).
  14. Schmid, P. J. Dynamic mode decomposition of numerical and experimental data. \JournalTitleJournal of fluid mechanics 656, 5–28 (2010).
  15. Dynamic mode decomposition with control. \JournalTitleSIAM Journal on Applied Dynamical Systems 15, 142–161 (2016).
  16. Discovering governing equations from data by sparse identification of nonlinear dynamical systems. \JournalTitleProceedings of the national academy of sciences 113, 3932–3937 (2016).
  17. Learning physics-based models from data: perspectives from inverse problems and model reduction. \JournalTitleActa Numerica 30, 445–554 (2021).
  18. Bayesian operator inference for data-driven reduced-order modeling. \JournalTitlearXiv preprint arXiv:2204.10829 (2022).
  19. Data-driven operator inference for nonintrusive projection-based model reduction. \JournalTitleComputer Methods in Applied Mechanics and Engineering 306, 196–215 (2016).
  20. Data-driven reduced order modeling for time-dependent problems. \JournalTitleComputer methods in applied mechanics and engineering 345, 75–99 (2019).
  21. Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders. \JournalTitleJournal of Computational Physics 404, 108973 (2020).
  22. Data-driven discovery of coordinates and governing equations. \JournalTitleProceedings of the National Academy of Sciences 116, 22445–22451 (2019).
  23. From pixels to torques: Policy learning with deep dynamical models. \JournalTitlearXiv preprint arXiv:1502.02251 (2015).
  24. Data-efficient learning of feedback policies from image pixels using deep dynamical models. \JournalTitlearXiv preprint arXiv:1510.02173 (2015).
  25. Auto-encoding variational bayes. \JournalTitlearXiv preprint arXiv:1312.6114 (2013).
  26. A disentangled recognition and nonlinear dynamics model for unsupervised learning. \JournalTitleAdvances in neural information processing systems 30 (2017).
  27. Deep kalman filters. \JournalTitlearXiv preprint arXiv:1511.05121 (2015).
  28. Deep variational bayes filters: Unsupervised learning of state space models from raw data. \JournalTitlearXiv preprint arXiv:1605.06432 (2016).
  29. Buesing, L. et al. Learning and querying fast generative models for reinforcement learning. \JournalTitlearXiv preprint arXiv:1802.03006 (2018).
  30. Doerr, A. et al. Probabilistic recurrent state-space models. In International Conference on Machine Learning, 1280–1289 (PMLR, 2018).
  31. Hafner, D. et al. Learning latent dynamics for planning from pixels. In International conference on machine learning, 2555–2565 (PMLR, 2019).
  32. Dream to control: Learning behaviors by latent imagination. \JournalTitlearXiv preprint arXiv:1912.01603 (2019).
  33. Deep kernel learning. In Artificial intelligence and statistics, 370–378 (PMLR, 2016).
  34. Gaussian processes for machine learning, vol. 2 (MIT press Cambridge, MA, 2006).
  35. Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems. \JournalTitleIEEE Transactions on Neural Networks 6, 911–917 (1995).
  36. Manifold gaussian processes for regression. In 2016 International Joint Conference on Neural Networks (IJCNN), 3338–3345 (IEEE, 2016).
  37. Adversarial examples, uncertainty, and transfer testing robustness in gaussian process hybrid deep networks. \JournalTitlearXiv preprint arXiv:1707.02476 (2017).
  38. Overview of deep kernel learning based techniques and applications. \JournalTitleJ. Netw. Intell. 1, 83–98 (2016).
  39. The promises and pitfalls of deep kernel learning. In Uncertainty in Artificial Intelligence, 1206–1216 (PMLR, 2021).
  40. Bridging deep and kernel methods. In ESANN2017: 25th European Symposium on Artificial Neural Networks: Bruges, Belgium, April 26-27-28, 1–10 (2017).
  41. Adaptive deep kernel learning. \JournalTitlearXiv preprint arXiv:1905.12131 (2019).
  42. Stochastic variational deep kernel learning. \JournalTitleAdvances in Neural Information Processing Systems 29 (2016).
  43. Kononenko, I. Bayesian neural networks. \JournalTitleBiological Cybernetics 61, 361–370 (1989).
  44. Mastering atari with discrete world models. \JournalTitlearXiv preprint arXiv:2010.02193 (2020).
  45. Visualizing data using t-sne. \JournalTitleJournal of machine learning research 9 (2008).
  46. Abdar, M. et al. A review of uncertainty quantification in deep learning: Techniques, applications and challenges. \JournalTitleInformation Fusion 76, 243–297 (2021).
  47. Settles, B. Active learning. \JournalTitleSynthesis lectures on artificial intelligence and machine learning 6, 1–114 (2012).
  48. Ensemble-sindy: Robust sparse model discovery in the low-data, high-noise limit, with active learning and control. \JournalTitleProceedings of the Royal Society A 478, 20210904 (2022).
  49. Gpytorch: Blackbox matrix-matrix gaussian process inference with gpu acceleration. \JournalTitleAdvances in neural information processing systems 31 (2018).
  50. Brockman, G. et al. Openai gym. \JournalTitlearXiv preprint arXiv:1606.01540 (2016).
Citations (9)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now