Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms

Mathieu Blondel, Masakazu Ishihata, Akinori Fujino, Naonori Ueda
Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:850-858, 2016.

Abstract

Polynomial networks and factorization machines are two recently-proposed models that can efficiently use feature interactions in classification and regression tasks. In this paper, we revisit both models from a unified perspective. Based on this new view, we study the properties of both models and propose new efficient training algorithms. Key to our approach is to cast parameter learning as a low-rank symmetric tensor estimation problem, which we solve by multi-convex optimization. We demonstrate our approach on regression and recommender system tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v48-blondel16, title = {Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms}, author = {Blondel, Mathieu and Ishihata, Masakazu and Fujino, Akinori and Ueda, Naonori}, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {850--858}, 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/blondel16.pdf}, url = {https://proceedings.mlr.press/v48/blondel16.html}, abstract = {Polynomial networks and factorization machines are two recently-proposed models that can efficiently use feature interactions in classification and regression tasks. In this paper, we revisit both models from a unified perspective. Based on this new view, we study the properties of both models and propose new efficient training algorithms. Key to our approach is to cast parameter learning as a low-rank symmetric tensor estimation problem, which we solve by multi-convex optimization. We demonstrate our approach on regression and recommender system tasks.} }
Endnote
%0 Conference Paper %T Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms %A Mathieu Blondel %A Masakazu Ishihata %A Akinori Fujino %A Naonori Ueda %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-blondel16 %I PMLR %P 850--858 %U https://proceedings.mlr.press/v48/blondel16.html %V 48 %X Polynomial networks and factorization machines are two recently-proposed models that can efficiently use feature interactions in classification and regression tasks. In this paper, we revisit both models from a unified perspective. Based on this new view, we study the properties of both models and propose new efficient training algorithms. Key to our approach is to cast parameter learning as a low-rank symmetric tensor estimation problem, which we solve by multi-convex optimization. We demonstrate our approach on regression and recommender system tasks.
RIS
TY - CPAPER TI - Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms AU - Mathieu Blondel AU - Masakazu Ishihata AU - Akinori Fujino AU - Naonori Ueda 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-blondel16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 48 SP - 850 EP - 858 L1 - http://proceedings.mlr.press/v48/blondel16.pdf UR - https://proceedings.mlr.press/v48/blondel16.html AB - Polynomial networks and factorization machines are two recently-proposed models that can efficiently use feature interactions in classification and regression tasks. In this paper, we revisit both models from a unified perspective. Based on this new view, we study the properties of both models and propose new efficient training algorithms. Key to our approach is to cast parameter learning as a low-rank symmetric tensor estimation problem, which we solve by multi-convex optimization. We demonstrate our approach on regression and recommender system tasks. ER -
APA
Blondel, M., Ishihata, M., Fujino, A. & Ueda, N.. (2016). Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms. Proceedings of The 33rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 48:850-858 Available from https://proceedings.mlr.press/v48/blondel16.html.

Related Material