ABSTRACT
The Gaussian process (GP) has been widely used to learn the dynamical system from training data. Current methods ignore the dependencies among the multiple dimensions of the system function and model each dimension of the system function using an independent single-output GP. This paper proposes a novel multiple-output Gaussian process model to learn the dynamical system. We assume that ordinary differential equations (ODEs) with unknown physical parameters are available as prior information. The system function is modeled by a GP with the mean function given by the one-step integration of ODEs. By this way, all dimensions of the system function are correlated because they share the same unknown physical parameters. This correlation allows the flow of information between multiple dimensions of the system function which in turn will lead to an improved model. With the existing models as baselines, we show the benefits of the proposed model by simulations.
- G. Pillonetto and G.D. Nicolao, "A new kernel-based approach for linear system identification," Automatica, vol.46, no.1, pp.81-93, 2010.Google ScholarDigital Library
- K. Fujimoto, A. Taniguchi and Y. Nishida, "System Identification of Nonlinear State-Space Models with Linearly Dependent Unknown Parameters Based on Variational Bayes," SICE Journal of Control, Measurement, and System Integration, vol.11, no.6, pp.456-462, 2018.Google ScholarCross Ref
- M.P. Deisenroth and C.E. Rasmussen, "PILCO: A model-based and data-efficient approach to policy search,” Proceedings of the 28th International Conference on machine learning, 2011.Google Scholar
- K. Fujimoto, H. Beppu and Y. Takaki, "Numerical Solutions of Hamilton-Jacobi Inequalities by Constrained Gaussian Process Regression," SICE Journal of Control, Measurement, and System Integration, vol.11, no.5, pp.419-428, 2018.Google ScholarCross Ref
- J. Ko, D.J. Klein, D. Fox and D. Haehnel, "Gaussian processes and reinforcement learning for identification and control of an autonomous blimp," Proceedings 2007 ieee international conference on robotics and automation, IEEE, pp.742-747, 2007.Google Scholar
- D. Romeres, M. Zorzi, R. Camoriano and A. Chiuso, "Online semi-parametric learning for inverse dynamics modeling," 2016 IEEE 55th Conference on Decision and Control (CDC), IEEE, pp.2945-2950, 2016.Google Scholar
- S. Tang, K. Fujimoto, and I. Maruta, "Learning Dynamic Systems Using Gaussian Process Regression with Analytic Ordinary Differential Equations as Prior Information,'' IEICE Transactions on Information and Systems, vol. 104, no. 9, pp. 1440–1449, 2021.Google ScholarCross Ref
- P. Goovaerts. Geostatistics for natural resources evaluation. Oxford University Press on Demand, 1997.Google Scholar
- M.A. Alvarez and N.D. Lawrence, "Computationally efficient convolved multiple output Gaussian processes,” The Journal of Machine Learning Research vol.12, pp.1459-1500, 2011.Google ScholarDigital Library
- C. Rasmussen, and C. Williams, Gaussian Processes for Machine Learning, The MIT Press, 2005.Google ScholarCross Ref
- E. Todorov, T. Erez, and Y. Tassa, "Mujoco: A physics engine for model-based control,'' 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 5026–5033, 2012.Google Scholar
Index Terms
- Learning dynamical systems using a novel multiple-output Gaussian process model
Recommendations
Gaussian Process Dynamical Autoencoder Model
ISMSI '19: Proceedings of the 2019 3rd International Conference on Intelligent Systems, Metaheuristics & Swarm IntelligenceDimension reduction realize extraction of substantial low dimensional latent structure in high-dimensional data. Due to recent developments in information and measurement technology, it becomes more important to develop dimension reduction algorithms ...
Analysing Dynamical Systems
SIMULTECH 2017: Proceedings of the 7th International Conference on Simulation and Modeling Methodologies, Technologies and ApplicationsOrdinary differential equations arise in a variety of applications, including e.g. climate systems, and can exhibit complicated dynamical behaviour. Complete Lyapunov functions can capture this behaviour by dividing the phase space into the chain-...
Variational dependent multi-output Gaussian process dynamical systems
This paper presents a dependent multi-output Gaussian process (GP) for modeling complex dynamical systems. The outputs are dependent in this model, which is largely different from previous GP dynamical systems. We adopt convolved multi-output GPs to ...
Comments