Non-Linear Cross-Domain Collaborative Filtering via Hyper-Structure Transfer

Yan-Fu Liu, Cheng-Yu Hsu, Shan-Hung Wu
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:1190-1198, 2015.

Abstract

The Cross Domain Collaborative Filtering (CDCF) exploits the rating matrices from multiple domains to make better recommendations. Existing CDCF methods adopt the sub-structure sharing technique that can only transfer linearly correlated knowledge between domains. In this paper, we propose the notion of Hyper-Structure Transfer (HST) that requires the rating matrices to be explained by the projections of some more complex structure, called the hyper-structure, shared by all domains, and thus allows the non-linearly correlated knowledge between domains to be identified and transferred. Extensive experiments are conducted and the results demonstrate the effectiveness of our HST models empirically.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-liua15, title = {Non-Linear Cross-Domain Collaborative Filtering via Hyper-Structure Transfer}, author = {Liu, Yan-Fu and Hsu, Cheng-Yu and Wu, Shan-Hung}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {1190--1198}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/liua15.pdf}, url = {https://proceedings.mlr.press/v37/liua15.html}, abstract = {The Cross Domain Collaborative Filtering (CDCF) exploits the rating matrices from multiple domains to make better recommendations. Existing CDCF methods adopt the sub-structure sharing technique that can only transfer linearly correlated knowledge between domains. In this paper, we propose the notion of Hyper-Structure Transfer (HST) that requires the rating matrices to be explained by the projections of some more complex structure, called the hyper-structure, shared by all domains, and thus allows the non-linearly correlated knowledge between domains to be identified and transferred. Extensive experiments are conducted and the results demonstrate the effectiveness of our HST models empirically.} }
Endnote
%0 Conference Paper %T Non-Linear Cross-Domain Collaborative Filtering via Hyper-Structure Transfer %A Yan-Fu Liu %A Cheng-Yu Hsu %A Shan-Hung Wu %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-liua15 %I PMLR %P 1190--1198 %U https://proceedings.mlr.press/v37/liua15.html %V 37 %X The Cross Domain Collaborative Filtering (CDCF) exploits the rating matrices from multiple domains to make better recommendations. Existing CDCF methods adopt the sub-structure sharing technique that can only transfer linearly correlated knowledge between domains. In this paper, we propose the notion of Hyper-Structure Transfer (HST) that requires the rating matrices to be explained by the projections of some more complex structure, called the hyper-structure, shared by all domains, and thus allows the non-linearly correlated knowledge between domains to be identified and transferred. Extensive experiments are conducted and the results demonstrate the effectiveness of our HST models empirically.
RIS
TY - CPAPER TI - Non-Linear Cross-Domain Collaborative Filtering via Hyper-Structure Transfer AU - Yan-Fu Liu AU - Cheng-Yu Hsu AU - Shan-Hung Wu BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-liua15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 1190 EP - 1198 L1 - http://proceedings.mlr.press/v37/liua15.pdf UR - https://proceedings.mlr.press/v37/liua15.html AB - The Cross Domain Collaborative Filtering (CDCF) exploits the rating matrices from multiple domains to make better recommendations. Existing CDCF methods adopt the sub-structure sharing technique that can only transfer linearly correlated knowledge between domains. In this paper, we propose the notion of Hyper-Structure Transfer (HST) that requires the rating matrices to be explained by the projections of some more complex structure, called the hyper-structure, shared by all domains, and thus allows the non-linearly correlated knowledge between domains to be identified and transferred. Extensive experiments are conducted and the results demonstrate the effectiveness of our HST models empirically. ER -
APA
Liu, Y., Hsu, C. & Wu, S.. (2015). Non-Linear Cross-Domain Collaborative Filtering via Hyper-Structure Transfer. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:1190-1198 Available from https://proceedings.mlr.press/v37/liua15.html.

Related Material