ACLS: Adaptive and Conditional Label Smoothing for Network Calibration
This work addresses network calibration for improving confidence estimates in deep learning models, presenting an incremental analysis and method.
The paper tackled the problem of network calibration in deep neural networks by analyzing existing regularization-based methods and introducing a new loss function called ACLS, which demonstrated effectiveness across multiple benchmarks including CIFAR10, Tiny-ImageNet, ImageNet, and PASCAL VOC.
We address the problem of network calibration adjusting miscalibrated confidences of deep neural networks. Many approaches to network calibration adopt a regularization-based method that exploits a regularization term to smooth the miscalibrated confidences. Although these approaches have shown the effectiveness on calibrating the networks, there is still a lack of understanding on the underlying principles of regularization in terms of network calibration. We present in this paper an in-depth analysis of existing regularization-based methods, providing a better understanding on how they affect to network calibration. Specifically, we have observed that 1) the regularization-based methods can be interpreted as variants of label smoothing, and 2) they do not always behave desirably. Based on the analysis, we introduce a novel loss function, dubbed ACLS, that unifies the merits of existing regularization methods, while avoiding the limitations. We show extensive experimental results for image classification and semantic segmentation on standard benchmarks, including CIFAR10, Tiny-ImageNet, ImageNet, and PASCAL VOC, demonstrating the effectiveness of our loss function.