Abstract
A Gaussian process model can be learned from data by identifying the covariance matrix of its sample values. The matrix usually depends on some fixed parameters called input length scales. Their estimation is equivalent to finding the corresponding diffeomorphism of the process inputs. Spatially variable length scales are difficult to estimate in the absence of good a priori values. We suggest a fluid-based nonlinear map of the process inputs and our experiments validate such a model on a synthetic problem.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Alberty, J., Carstensen, C., Funken, S.A., Klose, R.: Matlab Implementation of the Finite Element Method in Elasticity. Computing 69(3), 239–263 (2002)
Christensen, G.: Deformable Shape Models for Anatomy. Ph.d. thesis, Washington University (1994)
Gibbs, M.N.: Bayesian Gaussian Processes for Regression and Classification. Ph.d. thesis, Cambridge University (1997)
Grenander, U., Miller, M.I.: Computational Anatomy: An Emerging Discipline. Quart. Appl. Math. LVI(4), 617–694 (1998)
Kärkkäinen, T., Heikkola, E.: Robust Formulations for Training Multilayer Perceptrons. Neural Computation 16(4), 837–862 (2004)
Neal, R.M.: Priors for Infinite Networks. Technical Report CRG-TR-94-1, The University of Toronto (1994)
Paciorek, C.J.: Nonstationary Gaussian Processes for Regression and Spatial Modelling. Ph.d. thesis, Carnegie Mellon University (2003)
Rasmussen, C.E.: Evaluation of Gaussian Processes and Other Methods for Nonlinear Regression. Ph.d. thesis, The University of Toronto (1996)
Wollny, G., Kruggel, F.: Computational Cost of Nonrigid Registration Algorithms Based on Fluid Dynamics. IEEE Trans. on Medical Imaging 21(8), 946–952 (2002)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2004 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Girdziušas, R., Laaksonen, J. (2004). Gaussian Process Regression with Fluid Hyperpriors. In: Pal, N.R., Kasabov, N., Mudi, R.K., Pal, S., Parui, S.K. (eds) Neural Information Processing. ICONIP 2004. Lecture Notes in Computer Science, vol 3316. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-30499-9_87
Download citation
DOI: https://doi.org/10.1007/978-3-540-30499-9_87
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-23931-4
Online ISBN: 978-3-540-30499-9
eBook Packages: Springer Book Archive