Learning from Multiple Annotator Noisy Labels via Sample-wise Label Fusion
This addresses the challenge of noisy labeling in real-world applications where accurate annotations are not feasible, offering a more flexible approach for domains like image classification.
The paper tackles the problem of learning classifiers from datasets with multiple noisy labels per sample, proposing a method that models annotator errors as both annotator and sample dependent, achieving superior performance on MNIST, CIFAR-100, and ImageNet-100 compared to state-of-the-art baselines.
Data lies at the core of modern deep learning. The impressive performance of supervised learning is built upon a base of massive accurately labeled data. However, in some real-world applications, accurate labeling might not be viable; instead, multiple noisy labels (instead of one accurate label) are provided by several annotators for each data sample. Learning a classifier on such a noisy training dataset is a challenging task. Previous approaches usually assume that all data samples share the same set of parameters related to annotator errors, while we demonstrate that label error learning should be both annotator and data sample dependent. Motivated by this observation, we propose a novel learning algorithm. The proposed method displays superiority compared with several state-of-the-art baseline methods on MNIST, CIFAR-100, and ImageNet-100. Our code is available at: https://github.com/zhengqigao/Learning-from-Multiple-Annotator-Noisy-Labels.