Decouple then Classify: A Dynamic Multi-view Labeling Strategy with Shared and Specific Information

Xinhang Wan, Jiyuan Liu, Xinwang Liu, Yi Wen, Hao Yu, Siwei Wang, Shengju Yu, Tianjiao Wan, Jun Wang, En Zhu
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:49941-49956, 2024.

Abstract

Sample labeling is the most primary and fundamental step of semi-supervised learning. In literature, most existing methods randomly label samples with a given ratio, but achieve unpromising and unstable results due to the randomness, especially in multi-view settings. To address this issue, we propose a Dynamic Multi-view Labeling Strategy with Shared and Specific Information. To be brief, by building two classifiers with existing labels to utilize decoupled shared and specific information, we select the samples of low classification confidence and label them in high priorities. The newly generated labels are also integrated to update the classifiers adaptively. The two processes are executed alternatively until a satisfying classification performance. To validate the effectiveness of the proposed method, we conduct extensive experiments on popular benchmarks, achieving promising performance. The code is publicly available at https://github.com/wanxinhang/ICML2024_decouple_then_classify.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-wan24e, title = {Decouple then Classify: A Dynamic Multi-view Labeling Strategy with Shared and Specific Information}, author = {Wan, Xinhang and Liu, Jiyuan and Liu, Xinwang and Wen, Yi and Yu, Hao and Wang, Siwei and Yu, Shengju and Wan, Tianjiao and Wang, Jun and Zhu, En}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {49941--49956}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/wan24e/wan24e.pdf}, url = {https://proceedings.mlr.press/v235/wan24e.html}, abstract = {Sample labeling is the most primary and fundamental step of semi-supervised learning. In literature, most existing methods randomly label samples with a given ratio, but achieve unpromising and unstable results due to the randomness, especially in multi-view settings. To address this issue, we propose a Dynamic Multi-view Labeling Strategy with Shared and Specific Information. To be brief, by building two classifiers with existing labels to utilize decoupled shared and specific information, we select the samples of low classification confidence and label them in high priorities. The newly generated labels are also integrated to update the classifiers adaptively. The two processes are executed alternatively until a satisfying classification performance. To validate the effectiveness of the proposed method, we conduct extensive experiments on popular benchmarks, achieving promising performance. The code is publicly available at https://github.com/wanxinhang/ICML2024_decouple_then_classify.} }
Endnote
%0 Conference Paper %T Decouple then Classify: A Dynamic Multi-view Labeling Strategy with Shared and Specific Information %A Xinhang Wan %A Jiyuan Liu %A Xinwang Liu %A Yi Wen %A Hao Yu %A Siwei Wang %A Shengju Yu %A Tianjiao Wan %A Jun Wang %A En Zhu %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-wan24e %I PMLR %P 49941--49956 %U https://proceedings.mlr.press/v235/wan24e.html %V 235 %X Sample labeling is the most primary and fundamental step of semi-supervised learning. In literature, most existing methods randomly label samples with a given ratio, but achieve unpromising and unstable results due to the randomness, especially in multi-view settings. To address this issue, we propose a Dynamic Multi-view Labeling Strategy with Shared and Specific Information. To be brief, by building two classifiers with existing labels to utilize decoupled shared and specific information, we select the samples of low classification confidence and label them in high priorities. The newly generated labels are also integrated to update the classifiers adaptively. The two processes are executed alternatively until a satisfying classification performance. To validate the effectiveness of the proposed method, we conduct extensive experiments on popular benchmarks, achieving promising performance. The code is publicly available at https://github.com/wanxinhang/ICML2024_decouple_then_classify.
APA
Wan, X., Liu, J., Liu, X., Wen, Y., Yu, H., Wang, S., Yu, S., Wan, T., Wang, J. & Zhu, E.. (2024). Decouple then Classify: A Dynamic Multi-view Labeling Strategy with Shared and Specific Information. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:49941-49956 Available from https://proceedings.mlr.press/v235/wan24e.html.

Related Material