Maximum Margin Partial Label Learning

Fei Yu, Min-Ling Zhang
Asian Conference on Machine Learning, PMLR 45:96-111, 2016.

Abstract

Partial label learning deals with the problem that each training example is associated with a set of \emphcandidate labels, and only one among the set is the ground-truth label. The basic strategy to learn from partial label examples is disambiguation, i.e. by trying to recover the ground-truth labeling information from the candidate label set. As one of the major machine learning techniques, maximum margin criterion has been employed to solve the partial label learning problem. Therein, disambiguation is performed by optimizing the margin between the maximum modeling output from candidate labels and that from non-candidate labels. However, in this formulation the margin between the ground-truth label and other candidate labels is not differentiated. In this paper, a new maximum margin formulation for partial label learning is proposed which aims to directly maximize the margin between the ground-truth label and all other labels. Specifically, an alternating optimization procedure is utilized to coordinate \emphground-truth label identification and \emphmargin maximization. Extensive experiments show that the derived partial label learning approach achieves competitive performance against other state-of-the-art comparing approaches.

Cite this Paper


BibTeX
@InProceedings{pmlr-v45-Yu15, title = {Maximum Margin Partial Label Learning}, author = {Yu, Fei and Zhang, Min-Ling}, booktitle = {Asian Conference on Machine Learning}, pages = {96--111}, year = {2016}, editor = {Holmes, Geoffrey and Liu, Tie-Yan}, volume = {45}, series = {Proceedings of Machine Learning Research}, address = {Hong Kong}, month = {20--22 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v45/Yu15.pdf}, url = {https://proceedings.mlr.press/v45/Yu15.html}, abstract = {Partial label learning deals with the problem that each training example is associated with a set of \emphcandidate labels, and only one among the set is the ground-truth label. The basic strategy to learn from partial label examples is disambiguation, i.e. by trying to recover the ground-truth labeling information from the candidate label set. As one of the major machine learning techniques, maximum margin criterion has been employed to solve the partial label learning problem. Therein, disambiguation is performed by optimizing the margin between the maximum modeling output from candidate labels and that from non-candidate labels. However, in this formulation the margin between the ground-truth label and other candidate labels is not differentiated. In this paper, a new maximum margin formulation for partial label learning is proposed which aims to directly maximize the margin between the ground-truth label and all other labels. Specifically, an alternating optimization procedure is utilized to coordinate \emphground-truth label identification and \emphmargin maximization. Extensive experiments show that the derived partial label learning approach achieves competitive performance against other state-of-the-art comparing approaches. } }
Endnote
%0 Conference Paper %T Maximum Margin Partial Label Learning %A Fei Yu %A Min-Ling Zhang %B Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Geoffrey Holmes %E Tie-Yan Liu %F pmlr-v45-Yu15 %I PMLR %P 96--111 %U https://proceedings.mlr.press/v45/Yu15.html %V 45 %X Partial label learning deals with the problem that each training example is associated with a set of \emphcandidate labels, and only one among the set is the ground-truth label. The basic strategy to learn from partial label examples is disambiguation, i.e. by trying to recover the ground-truth labeling information from the candidate label set. As one of the major machine learning techniques, maximum margin criterion has been employed to solve the partial label learning problem. Therein, disambiguation is performed by optimizing the margin between the maximum modeling output from candidate labels and that from non-candidate labels. However, in this formulation the margin between the ground-truth label and other candidate labels is not differentiated. In this paper, a new maximum margin formulation for partial label learning is proposed which aims to directly maximize the margin between the ground-truth label and all other labels. Specifically, an alternating optimization procedure is utilized to coordinate \emphground-truth label identification and \emphmargin maximization. Extensive experiments show that the derived partial label learning approach achieves competitive performance against other state-of-the-art comparing approaches.
RIS
TY - CPAPER TI - Maximum Margin Partial Label Learning AU - Fei Yu AU - Min-Ling Zhang BT - Asian Conference on Machine Learning DA - 2016/02/25 ED - Geoffrey Holmes ED - Tie-Yan Liu ID - pmlr-v45-Yu15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 45 SP - 96 EP - 111 L1 - http://proceedings.mlr.press/v45/Yu15.pdf UR - https://proceedings.mlr.press/v45/Yu15.html AB - Partial label learning deals with the problem that each training example is associated with a set of \emphcandidate labels, and only one among the set is the ground-truth label. The basic strategy to learn from partial label examples is disambiguation, i.e. by trying to recover the ground-truth labeling information from the candidate label set. As one of the major machine learning techniques, maximum margin criterion has been employed to solve the partial label learning problem. Therein, disambiguation is performed by optimizing the margin between the maximum modeling output from candidate labels and that from non-candidate labels. However, in this formulation the margin between the ground-truth label and other candidate labels is not differentiated. In this paper, a new maximum margin formulation for partial label learning is proposed which aims to directly maximize the margin between the ground-truth label and all other labels. Specifically, an alternating optimization procedure is utilized to coordinate \emphground-truth label identification and \emphmargin maximization. Extensive experiments show that the derived partial label learning approach achieves competitive performance against other state-of-the-art comparing approaches. ER -
APA
Yu, F. & Zhang, M.. (2016). Maximum Margin Partial Label Learning. Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 45:96-111 Available from https://proceedings.mlr.press/v45/Yu15.html.

Related Material