Boolean Matrix Factorization and Noisy Completion via Message Passing

Siamak Ravanbakhsh, Barnabas Poczos, Russell Greiner
Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:945-954, 2016.

Abstract

Boolean matrix factorization and Boolean matrix completion from noisy observations are desirable unsupervised data-analysis methods due to their interpretability, but hard to perform due to their NP-hardness. We treat these problems as maximum a posteriori inference problems in a graphical model and present a message passing approach that scales linearly with the number of observations and factors. Our empirical study demonstrates that message passing is able to recover low-rank Boolean matrices, in the boundaries of theoretically possible recovery and compares favorably with state-of-the-art in real-world applications, such collaborative filtering with large-scale Boolean data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v48-ravanbakhsha16, title = {Boolean Matrix Factorization and Noisy Completion via Message Passing}, author = {Ravanbakhsh, Siamak and Poczos, Barnabas and Greiner, Russell}, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {945--954}, year = {2016}, editor = {Balcan, Maria Florina and Weinberger, Kilian Q.}, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/ravanbakhsha16.pdf}, url = {https://proceedings.mlr.press/v48/ravanbakhsha16.html}, abstract = {Boolean matrix factorization and Boolean matrix completion from noisy observations are desirable unsupervised data-analysis methods due to their interpretability, but hard to perform due to their NP-hardness. We treat these problems as maximum a posteriori inference problems in a graphical model and present a message passing approach that scales linearly with the number of observations and factors. Our empirical study demonstrates that message passing is able to recover low-rank Boolean matrices, in the boundaries of theoretically possible recovery and compares favorably with state-of-the-art in real-world applications, such collaborative filtering with large-scale Boolean data.} }
Endnote
%0 Conference Paper %T Boolean Matrix Factorization and Noisy Completion via Message Passing %A Siamak Ravanbakhsh %A Barnabas Poczos %A Russell Greiner %B Proceedings of The 33rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Maria Florina Balcan %E Kilian Q. Weinberger %F pmlr-v48-ravanbakhsha16 %I PMLR %P 945--954 %U https://proceedings.mlr.press/v48/ravanbakhsha16.html %V 48 %X Boolean matrix factorization and Boolean matrix completion from noisy observations are desirable unsupervised data-analysis methods due to their interpretability, but hard to perform due to their NP-hardness. We treat these problems as maximum a posteriori inference problems in a graphical model and present a message passing approach that scales linearly with the number of observations and factors. Our empirical study demonstrates that message passing is able to recover low-rank Boolean matrices, in the boundaries of theoretically possible recovery and compares favorably with state-of-the-art in real-world applications, such collaborative filtering with large-scale Boolean data.
RIS
TY - CPAPER TI - Boolean Matrix Factorization and Noisy Completion via Message Passing AU - Siamak Ravanbakhsh AU - Barnabas Poczos AU - Russell Greiner BT - Proceedings of The 33rd International Conference on Machine Learning DA - 2016/06/11 ED - Maria Florina Balcan ED - Kilian Q. Weinberger ID - pmlr-v48-ravanbakhsha16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 48 SP - 945 EP - 954 L1 - http://proceedings.mlr.press/v48/ravanbakhsha16.pdf UR - https://proceedings.mlr.press/v48/ravanbakhsha16.html AB - Boolean matrix factorization and Boolean matrix completion from noisy observations are desirable unsupervised data-analysis methods due to their interpretability, but hard to perform due to their NP-hardness. We treat these problems as maximum a posteriori inference problems in a graphical model and present a message passing approach that scales linearly with the number of observations and factors. Our empirical study demonstrates that message passing is able to recover low-rank Boolean matrices, in the boundaries of theoretically possible recovery and compares favorably with state-of-the-art in real-world applications, such collaborative filtering with large-scale Boolean data. ER -
APA
Ravanbakhsh, S., Poczos, B. & Greiner, R.. (2016). Boolean Matrix Factorization and Noisy Completion via Message Passing. Proceedings of The 33rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 48:945-954 Available from https://proceedings.mlr.press/v48/ravanbakhsha16.html.

Related Material