A Ranking-based KNN Approach for Multi-Label Classification

Tsung-Hsien Chiang, Hung-Yi Lo, Shou-De Lin
Proceedings of the Asian Conference on Machine Learning, PMLR 25:81-96, 2012.

Abstract

Multi-label classification has attracted a great deal of attention in recent years. This paper presents an interesting finding, namely, being able to identify neighbors with trustable labels can significantly improve the classification accuracy. Based on this finding, we propose a k-nearest-neighbor-based ranking approach to solve the multi-label classification problem. The approach exploits a ranking model to learn which neighbor’s labels are more trustable candidates for a weighted KNN-based strategy, and then assigns higher weights to those candidates when making weighted-voting decisions. The weights can then be determined by using a generalized pattern search technique. We collect several real-word data sets from various domains for the experiment. Our experiment results demonstrate that the proposed method outperforms state-of-the-art instance-based learning approaches. We believe that appropriately exploiting k-nearest neighbors is useful to solve the multi-label problem.

Cite this Paper

BibTeX
@InProceedings{pmlr-v25-chiang12, title = {A Ranking-based KNN Approach for Multi-Label Classification}, author = {Chiang, Tsung-Hsien and Lo, Hung-Yi and Lin, Shou-De}, booktitle = {Proceedings of the Asian Conference on Machine Learning}, pages = {81--96}, year = {2012}, editor = {Hoi, Steven C. H. and Buntine, Wray}, volume = {25}, series = {Proceedings of Machine Learning Research}, address = {Singapore Management University, Singapore}, month = {04--06 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v25/chiang12/chiang12.pdf}, url = {https://proceedings.mlr.press/v25/chiang12.html}, abstract = {Multi-label classification has attracted a great deal of attention in recent years. This paper presents an interesting finding, namely, being able to identify neighbors with trustable labels can significantly improve the classification accuracy. Based on this finding, we propose a k-nearest-neighbor-based ranking approach to solve the multi-label classification problem. The approach exploits a ranking model to learn which neighbor’s labels are more trustable candidates for a weighted KNN-based strategy, and then assigns higher weights to those candidates when making weighted-voting decisions. The weights can then be determined by using a generalized pattern search technique. We collect several real-word data sets from various domains for the experiment. Our experiment results demonstrate that the proposed method outperforms state-of-the-art instance-based learning approaches. We believe that appropriately exploiting k-nearest neighbors is useful to solve the multi-label problem.} }
Endnote
%0 Conference Paper %T A Ranking-based KNN Approach for Multi-Label Classification %A Tsung-Hsien Chiang %A Hung-Yi Lo %A Shou-De Lin %B Proceedings of the Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2012 %E Steven C. H. Hoi %E Wray Buntine %F pmlr-v25-chiang12 %I PMLR %P 81--96 %U https://proceedings.mlr.press/v25/chiang12.html %V 25 %X Multi-label classification has attracted a great deal of attention in recent years. This paper presents an interesting finding, namely, being able to identify neighbors with trustable labels can significantly improve the classification accuracy. Based on this finding, we propose a k-nearest-neighbor-based ranking approach to solve the multi-label classification problem. The approach exploits a ranking model to learn which neighbor’s labels are more trustable candidates for a weighted KNN-based strategy, and then assigns higher weights to those candidates when making weighted-voting decisions. The weights can then be determined by using a generalized pattern search technique. We collect several real-word data sets from various domains for the experiment. Our experiment results demonstrate that the proposed method outperforms state-of-the-art instance-based learning approaches. We believe that appropriately exploiting k-nearest neighbors is useful to solve the multi-label problem.
RIS
TY - CPAPER TI - A Ranking-based KNN Approach for Multi-Label Classification AU - Tsung-Hsien Chiang AU - Hung-Yi Lo AU - Shou-De Lin BT - Proceedings of the Asian Conference on Machine Learning DA - 2012/11/17 ED - Steven C. H. Hoi ED - Wray Buntine ID - pmlr-v25-chiang12 PB - PMLR DP - Proceedings of Machine Learning Research VL - 25 SP - 81 EP - 96 L1 - http://proceedings.mlr.press/v25/chiang12/chiang12.pdf UR - https://proceedings.mlr.press/v25/chiang12.html AB - Multi-label classification has attracted a great deal of attention in recent years. This paper presents an interesting finding, namely, being able to identify neighbors with trustable labels can significantly improve the classification accuracy. Based on this finding, we propose a k-nearest-neighbor-based ranking approach to solve the multi-label classification problem. The approach exploits a ranking model to learn which neighbor’s labels are more trustable candidates for a weighted KNN-based strategy, and then assigns higher weights to those candidates when making weighted-voting decisions. The weights can then be determined by using a generalized pattern search technique. We collect several real-word data sets from various domains for the experiment. Our experiment results demonstrate that the proposed method outperforms state-of-the-art instance-based learning approaches. We believe that appropriately exploiting k-nearest neighbors is useful to solve the multi-label problem. ER -
APA
Chiang, T., Lo, H. & Lin, S.. (2012). A Ranking-based KNN Approach for Multi-Label Classification. Proceedings of the Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 25:81-96 Available from https://proceedings.mlr.press/v25/chiang12.html.

Related Material