Magnitude-Preserving Ranking for Structured Outputs

Céline Brouard, Eric Bach, Sebastian Böcker, Juho Rousu
Proceedings of the Ninth Asian Conference on Machine Learning, PMLR 77:407-422, 2017.

Abstract

In this paper, we present a novel method for solving structured prediction problems, based on combining Input Output Kernel Regression (IOKR) with an extension of magnitude-preserving ranking to structured output spaces. In particular, we concentrate on the case where a set of candidate outputs has been given, and the associated pre-image problem calls for ranking the set of candidate outputs. Our method, called magnitude-preserving IOKR, both aims to produce a good approximation of the output feature vectors, and to preserve the magnitude differences of the output features in the candidate sets. For the case where the candidate set does not contain corresponding ’correct’ inputs, we propose a method for approximating the inputs through application of IOKR in the reverse direction. We apply our method to two learning problems: cross-lingual document retrieval and metabolite identification. Experiments show that the proposed approach improves performance over IOKR, and in the latter application obtains the current state-of-the-art accuracy.

Cite this Paper


BibTeX
@InProceedings{pmlr-v77-brouard17a, title = {Magnitude-Preserving Ranking for Structured Outputs}, author = {Brouard, Céline and Bach, Eric and Böcker, Sebastian and Rousu, Juho}, booktitle = {Proceedings of the Ninth Asian Conference on Machine Learning}, pages = {407--422}, year = {2017}, editor = {Zhang, Min-Ling and Noh, Yung-Kyun}, volume = {77}, series = {Proceedings of Machine Learning Research}, address = {Yonsei University, Seoul, Republic of Korea}, month = {15--17 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v77/brouard17a/brouard17a.pdf}, url = {https://proceedings.mlr.press/v77/brouard17a.html}, abstract = {In this paper, we present a novel method for solving structured prediction problems, based on combining Input Output Kernel Regression (IOKR) with an extension of magnitude-preserving ranking to structured output spaces. In particular, we concentrate on the case where a set of candidate outputs has been given, and the associated pre-image problem calls for ranking the set of candidate outputs. Our method, called magnitude-preserving IOKR, both aims to produce a good approximation of the output feature vectors, and to preserve the magnitude differences of the output features in the candidate sets. For the case where the candidate set does not contain corresponding ’correct’ inputs, we propose a method for approximating the inputs through application of IOKR in the reverse direction. We apply our method to two learning problems: cross-lingual document retrieval and metabolite identification. Experiments show that the proposed approach improves performance over IOKR, and in the latter application obtains the current state-of-the-art accuracy.} }
Endnote
%0 Conference Paper %T Magnitude-Preserving Ranking for Structured Outputs %A Céline Brouard %A Eric Bach %A Sebastian Böcker %A Juho Rousu %B Proceedings of the Ninth Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Min-Ling Zhang %E Yung-Kyun Noh %F pmlr-v77-brouard17a %I PMLR %P 407--422 %U https://proceedings.mlr.press/v77/brouard17a.html %V 77 %X In this paper, we present a novel method for solving structured prediction problems, based on combining Input Output Kernel Regression (IOKR) with an extension of magnitude-preserving ranking to structured output spaces. In particular, we concentrate on the case where a set of candidate outputs has been given, and the associated pre-image problem calls for ranking the set of candidate outputs. Our method, called magnitude-preserving IOKR, both aims to produce a good approximation of the output feature vectors, and to preserve the magnitude differences of the output features in the candidate sets. For the case where the candidate set does not contain corresponding ’correct’ inputs, we propose a method for approximating the inputs through application of IOKR in the reverse direction. We apply our method to two learning problems: cross-lingual document retrieval and metabolite identification. Experiments show that the proposed approach improves performance over IOKR, and in the latter application obtains the current state-of-the-art accuracy.
APA
Brouard, C., Bach, E., Böcker, S. & Rousu, J.. (2017). Magnitude-Preserving Ranking for Structured Outputs. Proceedings of the Ninth Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 77:407-422 Available from https://proceedings.mlr.press/v77/brouard17a.html.

Related Material