Research
Print page Print page
Switch language
The Capital Region of Denmark - a part of Copenhagen University Hospital
Published

Variational Bayesian partially observed non-negative tensor factorization

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Harvard

Hinrich, JL, Nielsen, SFV, Madsen, KH & Morup, M 2018, Variational Bayesian partially observed non-negative tensor factorization. in 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. vol. 2018-September, 8516924, IEEE Computer Society Press, pp. 1-6, 28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018, Aalborg, Denmark, 17/09/2018. https://doi.org/10.1109/MLSP.2018.8516924

APA

Hinrich, J. L., Nielsen, S. F. V., Madsen, K. H., & Morup, M. (2018). Variational Bayesian partially observed non-negative tensor factorization. In 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings (Vol. 2018-September, pp. 1-6). [8516924] IEEE Computer Society Press. https://doi.org/10.1109/MLSP.2018.8516924

CBE

Hinrich JL, Nielsen SFV, Madsen KH, Morup M. 2018. Variational Bayesian partially observed non-negative tensor factorization. In 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. IEEE Computer Society Press. pp. 1-6. https://doi.org/10.1109/MLSP.2018.8516924

MLA

Hinrich, Jesper L. et al. "Variational Bayesian partially observed non-negative tensor factorization". 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. IEEE Computer Society Press. 2018, 1-6. https://doi.org/10.1109/MLSP.2018.8516924

Vancouver

Hinrich JL, Nielsen SFV, Madsen KH, Morup M. Variational Bayesian partially observed non-negative tensor factorization. In 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. Vol. 2018-September. IEEE Computer Society Press. 2018. p. 1-6. 8516924 https://doi.org/10.1109/MLSP.2018.8516924

Author

Hinrich, Jesper L. ; Nielsen, Soren F.V. ; Madsen, Kristoffer H. ; Morup, Morten. / Variational Bayesian partially observed non-negative tensor factorization. 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. Vol. 2018-September IEEE Computer Society Press, 2018. pp. 1-6

Bibtex

@inproceedings{247e91e4213b4aed81577006d0741c6d,
title = "Variational Bayesian partially observed non-negative tensor factorization",
abstract = "Non-negative matrix and tensor factorization (NMF/NTF) have become important tools for extracting part based representations in data. It is however unclear when an NMF or NTF approach is most suited for data and how reliably the models predict when trained on partially observed data. We presently extend a recently proposed variational Bayesian NMF (VB-NMF) to non-negative tensor factorization (VB-NTF) for partially observed data. This admits bi- and multi-linear structure quantification considering both model prediction and evidence. We evaluate the developed VB-NTF on synthetic and a real dataset of gene expression in the human brain and contrast the performance to VB-NMF and conventional NMF/NTF. We find that the gene expressions are better accounted for by VB-NMF than VB-NTF and that VB-NMF/VB-NTF more robustly handle partially observed data than conventional NMF/NTF. In particular, probabilistic modeling is beneficial when large amounts of data is missing and/or the model order over-specified.",
keywords = "Human brain microarray data, Missing data, Non-negative tensor factorization, Probabilistic modeling",
author = "Hinrich, {Jesper L.} and Nielsen, {Soren F.V.} and Madsen, {Kristoffer H.} and Morten Morup",
year = "2018",
doi = "10.1109/MLSP.2018.8516924",
language = "English",
volume = "2018-September",
pages = "1--6",
booktitle = "2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings",
publisher = "IEEE Computer Society Press",
address = "United States",
note = "28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 ; Conference date: 17-09-2018 Through 20-09-2018",

}

RIS

TY - GEN

T1 - Variational Bayesian partially observed non-negative tensor factorization

AU - Hinrich, Jesper L.

AU - Nielsen, Soren F.V.

AU - Madsen, Kristoffer H.

AU - Morup, Morten

PY - 2018

Y1 - 2018

N2 - Non-negative matrix and tensor factorization (NMF/NTF) have become important tools for extracting part based representations in data. It is however unclear when an NMF or NTF approach is most suited for data and how reliably the models predict when trained on partially observed data. We presently extend a recently proposed variational Bayesian NMF (VB-NMF) to non-negative tensor factorization (VB-NTF) for partially observed data. This admits bi- and multi-linear structure quantification considering both model prediction and evidence. We evaluate the developed VB-NTF on synthetic and a real dataset of gene expression in the human brain and contrast the performance to VB-NMF and conventional NMF/NTF. We find that the gene expressions are better accounted for by VB-NMF than VB-NTF and that VB-NMF/VB-NTF more robustly handle partially observed data than conventional NMF/NTF. In particular, probabilistic modeling is beneficial when large amounts of data is missing and/or the model order over-specified.

AB - Non-negative matrix and tensor factorization (NMF/NTF) have become important tools for extracting part based representations in data. It is however unclear when an NMF or NTF approach is most suited for data and how reliably the models predict when trained on partially observed data. We presently extend a recently proposed variational Bayesian NMF (VB-NMF) to non-negative tensor factorization (VB-NTF) for partially observed data. This admits bi- and multi-linear structure quantification considering both model prediction and evidence. We evaluate the developed VB-NTF on synthetic and a real dataset of gene expression in the human brain and contrast the performance to VB-NMF and conventional NMF/NTF. We find that the gene expressions are better accounted for by VB-NMF than VB-NTF and that VB-NMF/VB-NTF more robustly handle partially observed data than conventional NMF/NTF. In particular, probabilistic modeling is beneficial when large amounts of data is missing and/or the model order over-specified.

KW - Human brain microarray data

KW - Missing data

KW - Non-negative tensor factorization

KW - Probabilistic modeling

UR - http://www.scopus.com/inward/record.url?scp=85057018485&partnerID=8YFLogxK

U2 - 10.1109/MLSP.2018.8516924

DO - 10.1109/MLSP.2018.8516924

M3 - Article in proceedings

AN - SCOPUS:85057018485

VL - 2018-September

SP - 1

EP - 6

BT - 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings

PB - IEEE Computer Society Press

T2 - 28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018

Y2 - 17 September 2018 through 20 September 2018

ER -

ID: 56438196