Partial Label Learning via Label Influence Function

Xiuwen Gong, Dong Yuan, Wei Bao
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:7665-7678, 2022.

Abstract

To deal with ambiguities in partial label learning (PLL), state-of-the-art strategies implement disambiguations by identifying the ground-truth label directly from the candidate label set. However, these approaches usually take the label that incurs a minimal loss as the ground-truth label or use the weight to represent which label has a high likelihood to be the ground-truth label. Little work has been done to investigate from the perspective of how a candidate label changing a predictive model. In this paper, inspired by influence function, we develop a novel PLL framework called Partial Label Learning via Label Influence Function (PLL-IF). Moreover, we implement the framework with two specific representative models, an SVM model and a neural network model, which are called PLL-IF+SVM and PLL-IF+NN method respectively. Extensive experiments conducted on various datasets demonstrate the superiorities of the proposed methods in terms of prediction accuracy, which in turn validates the effectiveness of the proposed PLL-IF framework.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-gong22c, title = {Partial Label Learning via Label Influence Function}, author = {Gong, Xiuwen and Yuan, Dong and Bao, Wei}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {7665--7678}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/gong22c/gong22c.pdf}, url = {https://proceedings.mlr.press/v162/gong22c.html}, abstract = {To deal with ambiguities in partial label learning (PLL), state-of-the-art strategies implement disambiguations by identifying the ground-truth label directly from the candidate label set. However, these approaches usually take the label that incurs a minimal loss as the ground-truth label or use the weight to represent which label has a high likelihood to be the ground-truth label. Little work has been done to investigate from the perspective of how a candidate label changing a predictive model. In this paper, inspired by influence function, we develop a novel PLL framework called Partial Label Learning via Label Influence Function (PLL-IF). Moreover, we implement the framework with two specific representative models, an SVM model and a neural network model, which are called PLL-IF+SVM and PLL-IF+NN method respectively. Extensive experiments conducted on various datasets demonstrate the superiorities of the proposed methods in terms of prediction accuracy, which in turn validates the effectiveness of the proposed PLL-IF framework.} }
Endnote
%0 Conference Paper %T Partial Label Learning via Label Influence Function %A Xiuwen Gong %A Dong Yuan %A Wei Bao %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-gong22c %I PMLR %P 7665--7678 %U https://proceedings.mlr.press/v162/gong22c.html %V 162 %X To deal with ambiguities in partial label learning (PLL), state-of-the-art strategies implement disambiguations by identifying the ground-truth label directly from the candidate label set. However, these approaches usually take the label that incurs a minimal loss as the ground-truth label or use the weight to represent which label has a high likelihood to be the ground-truth label. Little work has been done to investigate from the perspective of how a candidate label changing a predictive model. In this paper, inspired by influence function, we develop a novel PLL framework called Partial Label Learning via Label Influence Function (PLL-IF). Moreover, we implement the framework with two specific representative models, an SVM model and a neural network model, which are called PLL-IF+SVM and PLL-IF+NN method respectively. Extensive experiments conducted on various datasets demonstrate the superiorities of the proposed methods in terms of prediction accuracy, which in turn validates the effectiveness of the proposed PLL-IF framework.
APA
Gong, X., Yuan, D. & Bao, W.. (2022). Partial Label Learning via Label Influence Function. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:7665-7678 Available from https://proceedings.mlr.press/v162/gong22c.html.

Related Material