Multilabel Classification through Random Graph Ensembles

Hongyu Su, Juho Rousu
Proceedings of the 5th Asian Conference on Machine Learning, PMLR 29:404-418, 2013.

Abstract

We present new methods for multilabel classification, relying on ensemble learning on a collection of random output graphs imposed on the multilabel and a kernel-based structured output learner as the base classifier. For ensemble learning, differences among the output graphs provide the required base classifier diversity and lead to improved performance in the increasing size of the ensemble. We study different methods of forming the ensemble prediction, including majority voting and two methods that perform inferences over the graph structures before or after combining the base models into the ensemble. We compare the methods against the state-of-the-art machine learning approaches on a set of heterogeneous multilabel benchmark problems, including multilabel AdaBoost, convex multitask feature learning, as well as single target learning approaches represented by Bagging and SVM. In our experiments, the random graph ensembles are very competitive and robust, ranking first or second on most of the datasets. Overall, our results show that random graph ensembles are viable alternatives to flat multilabel and multitask learners.

Cite this Paper

BibTeX
@InProceedings{pmlr-v29-Su13, title = {Multilabel Classification through Random Graph Ensembles}, author = {Su, Hongyu and Rousu, Juho}, booktitle = {Proceedings of the 5th Asian Conference on Machine Learning}, pages = {404--418}, year = {2013}, editor = {Ong, Cheng Soon and Ho, Tu Bao}, volume = {29}, series = {Proceedings of Machine Learning Research}, address = {Australian National University, Canberra, Australia}, month = {13--15 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v29/Su13.pdf}, url = {https://proceedings.mlr.press/v29/Su13.html}, abstract = {We present new methods for multilabel classification, relying on ensemble learning on a collection of random output graphs imposed on the multilabel and a kernel-based structured output learner as the base classifier. For ensemble learning, differences among the output graphs provide the required base classifier diversity and lead to improved performance in the increasing size of the ensemble. We study different methods of forming the ensemble prediction, including majority voting and two methods that perform inferences over the graph structures before or after combining the base models into the ensemble. We compare the methods against the state-of-the-art machine learning approaches on a set of heterogeneous multilabel benchmark problems, including multilabel AdaBoost, convex multitask feature learning, as well as single target learning approaches represented by Bagging and SVM. In our experiments, the random graph ensembles are very competitive and robust, ranking first or second on most of the datasets. Overall, our results show that random graph ensembles are viable alternatives to flat multilabel and multitask learners.} }
Endnote
%0 Conference Paper %T Multilabel Classification through Random Graph Ensembles %A Hongyu Su %A Juho Rousu %B Proceedings of the 5th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2013 %E Cheng Soon Ong %E Tu Bao Ho %F pmlr-v29-Su13 %I PMLR %P 404--418 %U https://proceedings.mlr.press/v29/Su13.html %V 29 %X We present new methods for multilabel classification, relying on ensemble learning on a collection of random output graphs imposed on the multilabel and a kernel-based structured output learner as the base classifier. For ensemble learning, differences among the output graphs provide the required base classifier diversity and lead to improved performance in the increasing size of the ensemble. We study different methods of forming the ensemble prediction, including majority voting and two methods that perform inferences over the graph structures before or after combining the base models into the ensemble. We compare the methods against the state-of-the-art machine learning approaches on a set of heterogeneous multilabel benchmark problems, including multilabel AdaBoost, convex multitask feature learning, as well as single target learning approaches represented by Bagging and SVM. In our experiments, the random graph ensembles are very competitive and robust, ranking first or second on most of the datasets. Overall, our results show that random graph ensembles are viable alternatives to flat multilabel and multitask learners.
RIS
TY - CPAPER TI - Multilabel Classification through Random Graph Ensembles AU - Hongyu Su AU - Juho Rousu BT - Proceedings of the 5th Asian Conference on Machine Learning DA - 2013/10/21 ED - Cheng Soon Ong ED - Tu Bao Ho ID - pmlr-v29-Su13 PB - PMLR DP - Proceedings of Machine Learning Research VL - 29 SP - 404 EP - 418 L1 - http://proceedings.mlr.press/v29/Su13.pdf UR - https://proceedings.mlr.press/v29/Su13.html AB - We present new methods for multilabel classification, relying on ensemble learning on a collection of random output graphs imposed on the multilabel and a kernel-based structured output learner as the base classifier. For ensemble learning, differences among the output graphs provide the required base classifier diversity and lead to improved performance in the increasing size of the ensemble. We study different methods of forming the ensemble prediction, including majority voting and two methods that perform inferences over the graph structures before or after combining the base models into the ensemble. We compare the methods against the state-of-the-art machine learning approaches on a set of heterogeneous multilabel benchmark problems, including multilabel AdaBoost, convex multitask feature learning, as well as single target learning approaches represented by Bagging and SVM. In our experiments, the random graph ensembles are very competitive and robust, ranking first or second on most of the datasets. Overall, our results show that random graph ensembles are viable alternatives to flat multilabel and multitask learners. ER -
APA
Su, H. & Rousu, J.. (2013). Multilabel Classification through Random Graph Ensembles. Proceedings of the 5th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 29:404-418 Available from https://proceedings.mlr.press/v29/Su13.html.

Related Material