Delving Deep into Label Smoothing
This work provides an incremental improvement to label smoothing, a regularization technique, for researchers and practitioners working with deep neural networks to improve classification performance and robustness.
This paper proposes an Online Label Smoothing (OLS) strategy that dynamically generates soft labels based on model prediction statistics for the target category. This approach improves classification performance on CIFAR-100, ImageNet, and fine-grained datasets, and significantly enhances robustness to noisy labels compared to existing label smoothing methods.
Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches.