Efficient Learning of Mahalanobis Metrics for Ranking

Daryl Lim, Gert Lanckriet
Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):1980-1988, 2014.

Abstract

We develop an efficient algorithm to learn a Mahalanobis distance metric by directly optimizing a ranking loss. Our approach focuses on optimizing the top of the induced rankings, which is desirable in tasks such as visualization and nearest-neighbor retrieval. We further develop and justify a simple technique to reduce training time significantly with minimal impact on performance. Our proposed method significantly outperforms alternative methods on several real-world tasks, and can scale to large and high-dimensional data.

Cite this Paper


BibTeX
@InProceedings{pmlr-v32-lim14, title = {Efficient Learning of Mahalanobis Metrics for Ranking}, author = {Lim, Daryl and Lanckriet, Gert}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {1980--1988}, year = {2014}, editor = {Xing, Eric P. and Jebara, Tony}, volume = {32}, number = {2}, series = {Proceedings of Machine Learning Research}, address = {Bejing, China}, month = {22--24 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v32/lim14.pdf}, url = {https://proceedings.mlr.press/v32/lim14.html}, abstract = {We develop an efficient algorithm to learn a Mahalanobis distance metric by directly optimizing a ranking loss. Our approach focuses on optimizing the top of the induced rankings, which is desirable in tasks such as visualization and nearest-neighbor retrieval. We further develop and justify a simple technique to reduce training time significantly with minimal impact on performance. Our proposed method significantly outperforms alternative methods on several real-world tasks, and can scale to large and high-dimensional data.} }
Endnote
%0 Conference Paper %T Efficient Learning of Mahalanobis Metrics for Ranking %A Daryl Lim %A Gert Lanckriet %B Proceedings of the 31st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2014 %E Eric P. Xing %E Tony Jebara %F pmlr-v32-lim14 %I PMLR %P 1980--1988 %U https://proceedings.mlr.press/v32/lim14.html %V 32 %N 2 %X We develop an efficient algorithm to learn a Mahalanobis distance metric by directly optimizing a ranking loss. Our approach focuses on optimizing the top of the induced rankings, which is desirable in tasks such as visualization and nearest-neighbor retrieval. We further develop and justify a simple technique to reduce training time significantly with minimal impact on performance. Our proposed method significantly outperforms alternative methods on several real-world tasks, and can scale to large and high-dimensional data.
RIS
TY - CPAPER TI - Efficient Learning of Mahalanobis Metrics for Ranking AU - Daryl Lim AU - Gert Lanckriet BT - Proceedings of the 31st International Conference on Machine Learning DA - 2014/06/18 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-lim14 PB - PMLR DP - Proceedings of Machine Learning Research VL - 32 IS - 2 SP - 1980 EP - 1988 L1 - http://proceedings.mlr.press/v32/lim14.pdf UR - https://proceedings.mlr.press/v32/lim14.html AB - We develop an efficient algorithm to learn a Mahalanobis distance metric by directly optimizing a ranking loss. Our approach focuses on optimizing the top of the induced rankings, which is desirable in tasks such as visualization and nearest-neighbor retrieval. We further develop and justify a simple technique to reduce training time significantly with minimal impact on performance. Our proposed method significantly outperforms alternative methods on several real-world tasks, and can scale to large and high-dimensional data. ER -
APA
Lim, D. & Lanckriet, G.. (2014). Efficient Learning of Mahalanobis Metrics for Ranking. Proceedings of the 31st International Conference on Machine Learning, in Proceedings of Machine Learning Research 32(2):1980-1988 Available from https://proceedings.mlr.press/v32/lim14.html.

Related Material