Learning the Structure of Sum-Product Networks via an SVD-based Algorithm

Tameem Adel Radboud University Nijmegen, David Balduzzi Victoria University of Wellington, Ali Ghodsi
Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence, PMLR R13:219-228, 2015.

Abstract

Sum-product networks (SPNs) are a recently developed class of deep probabilistic models where inference is tractable. We present two new structure learning algorithms for sum-product networks, in the generative and discriminative settings, that are based on recursively extracting rank-one submatrices from data. The proposed algorithms find the subSPNs that are the most coherent jointly in the instances and variables – that is, whose instances are most strongly correlated over the given variables. Experimental results show that SPNs learned using the proposed generative algorithm have better likelihood and inference results – is also much faster than – than previous approaches. Finally, we apply the discriminative SPN structure learning algorithm to handwritten digit recognition tasks, where it achieves state-of-the-art performance for an SPN.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR13-nijmegen15a, title = {Learning the Structure of Sum-Product Networks via an {SVD}-based Algorithm}, author = {Nijmegen, Tameem Adel Radboud University and Wellington, David Balduzzi Victoria University of and Ghodsi, Ali}, booktitle = {Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence}, pages = {219--228}, year = {2015}, editor = {Meila, Marina and Heskes, Tom}, volume = {R13}, series = {Proceedings of Machine Learning Research}, month = {12--16 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r13/main/assets/nijmegen15a/nijmegen15a.pdf}, url = {https://proceedings.mlr.press/r13/nijmegen15a.html}, abstract = {Sum-product networks (SPNs) are a recently developed class of deep probabilistic models where inference is tractable. We present two new structure learning algorithms for sum-product networks, in the generative and discriminative settings, that are based on recursively extracting rank-one submatrices from data. The proposed algorithms find the subSPNs that are the most coherent jointly in the instances and variables – that is, whose instances are most strongly correlated over the given variables. Experimental results show that SPNs learned using the proposed generative algorithm have better likelihood and inference results – is also much faster than – than previous approaches. Finally, we apply the discriminative SPN structure learning algorithm to handwritten digit recognition tasks, where it achieves state-of-the-art performance for an SPN.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Learning the Structure of Sum-Product Networks via an SVD-based Algorithm %A Tameem Adel Radboud University Nijmegen %A David Balduzzi Victoria University of Wellington %A Ali Ghodsi %B Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2015 %E Marina Meila %E Tom Heskes %F pmlr-vR13-nijmegen15a %I PMLR %P 219--228 %U https://proceedings.mlr.press/r13/nijmegen15a.html %V R13 %X Sum-product networks (SPNs) are a recently developed class of deep probabilistic models where inference is tractable. We present two new structure learning algorithms for sum-product networks, in the generative and discriminative settings, that are based on recursively extracting rank-one submatrices from data. The proposed algorithms find the subSPNs that are the most coherent jointly in the instances and variables – that is, whose instances are most strongly correlated over the given variables. Experimental results show that SPNs learned using the proposed generative algorithm have better likelihood and inference results – is also much faster than – than previous approaches. Finally, we apply the discriminative SPN structure learning algorithm to handwritten digit recognition tasks, where it achieves state-of-the-art performance for an SPN. %Z Reissued by PMLR on 04 October 2026.
APA
Nijmegen, T.A.R.U., Wellington, D.B.V.U.o. & Ghodsi, A.. (2015). Learning the Structure of Sum-Product Networks via an SVD-based Algorithm. Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R13:219-228 Available from https://proceedings.mlr.press/r13/nijmegen15a.html. Reissued by PMLR on 04 October 2026.

Related Material