Interpretation and Generalization of Score Matching

Siwei Lyu
Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence, PMLR R7:367-374, 2009.

Abstract

Score matching is a recently developed parameter learning method that is particularly effective to complicated high dimensional density models with intractable partition functions. In this paper, we study two issues that have not been completely resolved for score matching. First, we provide a formal link between maximum likelihood and score matching. Our analysis shows that score matching finds model parameters that are more robust with noisy training data. Second, we develop a generalization of score matching. Based on this generalization, we further demonstrate an extension of score matching to models of discrete data.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR7-lyu09a, title = {Interpretation and Generalization of Score Matching}, author = {Lyu, Siwei}, booktitle = {Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence}, pages = {367--374}, year = {2009}, editor = {Bilmes, Jeff and Ng, Andrew Y.}, volume = {R7}, series = {Proceedings of Machine Learning Research}, month = {18--21 Jun}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r7/main/assets/lyu09a/lyu09a.pdf}, url = {https://proceedings.mlr.press/r7/lyu09a.html}, abstract = {Score matching is a recently developed parameter learning method that is particularly effective to complicated high dimensional density models with intractable partition functions. In this paper, we study two issues that have not been completely resolved for score matching. First, we provide a formal link between maximum likelihood and score matching. Our analysis shows that score matching finds model parameters that are more robust with noisy training data. Second, we develop a generalization of score matching. Based on this generalization, we further demonstrate an extension of score matching to models of discrete data.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Interpretation and Generalization of Score Matching %A Siwei Lyu %B Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2009 %E Jeff Bilmes %E Andrew Y. Ng %F pmlr-vR7-lyu09a %I PMLR %P 367--374 %U https://proceedings.mlr.press/r7/lyu09a.html %V R7 %X Score matching is a recently developed parameter learning method that is particularly effective to complicated high dimensional density models with intractable partition functions. In this paper, we study two issues that have not been completely resolved for score matching. First, we provide a formal link between maximum likelihood and score matching. Our analysis shows that score matching finds model parameters that are more robust with noisy training data. Second, we develop a generalization of score matching. Based on this generalization, we further demonstrate an extension of score matching to models of discrete data. %Z Reissued by PMLR on 04 October 2026.
APA
Lyu, S.. (2009). Interpretation and Generalization of Score Matching. Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R7:367-374 Available from https://proceedings.mlr.press/r7/lyu09a.html. Reissued by PMLR on 04 October 2026.

Related Material