Learning with Noisy Labels: Interconnection of Two Expectation-Maximizations
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911751618953216 |
|---|---|
| author | Kim, Heewon Chang, Hyun Sung Cho, Kiho Lee, Jaeyun Han, Bohyung |
| author_facet | Kim, Heewon Chang, Hyun Sung Cho, Kiho Lee, Jaeyun Han, Bohyung |
| contents | Labor-intensive labeling becomes a bottleneck in developing computer vision algorithms based on deep learning. For this reason, dealing with imperfect labels has increasingly gained attention and has become an active field of study. We address learning with noisy labels (LNL) problem, which is formalized as a task of finding a structured manifold in the midst of noisy data. In this framework, we provide a proper objective function and an optimization algorithm based on two expectation-maximization (EM) cycles. The separate networks associated with the two EM cycles collaborate to optimize the objective function, where one model is for distinguishing clean labels from corrupted ones while the other is for refurbishing the corrupted labels. This approach results in a non-collapsing LNL-flywheel model in the end. Experiments show that our algorithm achieves state-of-the-art performance in multiple standard benchmarks with substantial margins under various types of label noise. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_04390 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning with Noisy Labels: Interconnection of Two Expectation-Maximizations Kim, Heewon Chang, Hyun Sung Cho, Kiho Lee, Jaeyun Han, Bohyung Computer Vision and Pattern Recognition Labor-intensive labeling becomes a bottleneck in developing computer vision algorithms based on deep learning. For this reason, dealing with imperfect labels has increasingly gained attention and has become an active field of study. We address learning with noisy labels (LNL) problem, which is formalized as a task of finding a structured manifold in the midst of noisy data. In this framework, we provide a proper objective function and an optimization algorithm based on two expectation-maximization (EM) cycles. The separate networks associated with the two EM cycles collaborate to optimize the objective function, where one model is for distinguishing clean labels from corrupted ones while the other is for refurbishing the corrupted labels. This approach results in a non-collapsing LNL-flywheel model in the end. Experiments show that our algorithm achieves state-of-the-art performance in multiple standard benchmarks with substantial margins under various types of label noise. |
| title | Learning with Noisy Labels: Interconnection of Two Expectation-Maximizations |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2401.04390 |