Abstract
Statistical models are often based on non-normalized probability densities. That is, the model contains an unknown normalization constant whose computation is too difficult for practical purposes. Such models were encountered, for example, in Sects. 13.1.5 and 13.1.7. Maximum likelihood estimation is not possible without computation of the normalization constant. In this chapter, we show how such models can be estimated using a different estimation method. It is not necessary to know this material to understand the developments in this book; this is meant as supplementary material.
This chapter is based on (Hyvärinen 2005), first published in Journal of Machine Learning Research. Copyright retained by the author
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.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2009 Springer-Verlag London Limited
About this chapter
Cite this chapter
Hyvärinen, A., Hurri, J., Hoyer, P.O. (2009). Estimation of Non-normalized Statistical Models. In: Natural Image Statistics. Computational Imaging and Vision, vol 39. Springer, London. https://doi.org/10.1007/978-1-84882-491-1_21
Download citation
DOI: https://doi.org/10.1007/978-1-84882-491-1_21
Publisher Name: Springer, London
Print ISBN: 978-1-84882-490-4
Online ISBN: 978-1-84882-491-1
eBook Packages: Computer ScienceComputer Science (R0)