Dual-Branch Temperature Scaling Calibration for Long-Tailed Recognition

TL;DR

This paper introduces Dual-TS, a dual-branch temperature scaling calibration method designed for long-tailed recognition, addressing miscalibration issues especially in minority classes, and proposes a new evaluation metric Esbin-ECE.

Contribution

The paper proposes a novel dual-branch temperature scaling calibration model and a new calibration evaluation metric tailored for long-tailed data distributions.

Findings

Achieves state-of-the-art calibration performance on long-tailed datasets.

Effectively calibrates confidence levels across both majority and minority classes.

Improves evaluation accuracy with the Esbin-ECE metric.

Abstract

The calibration for deep neural networks is currently receiving widespread attention and research. Miscalibration usually leads to overconfidence of the model. While, under the condition of long-tailed distribution of data, the problem of miscalibration is more prominent due to the different confidence levels of samples in minority and majority categories, and it will result in more serious overconfidence. To address this problem, some current research have designed diverse temperature coefficients for different categories based on temperature scaling (TS) method. However, in the case of rare samples in minority classes, the temperature coefficient is not generalizable, and there is a large difference between the temperature coefficients of the training set and the validation set. To solve this challenge, this paper proposes a dual-branch temperature scaling calibration model (Dual-TS), which considers the diversities in temperature parameters of different categories and the non-generalizability of temperature parameters for rare samples in minority classes simultaneously. Moreover, we noticed that the traditional calibration evaluation metric, Excepted Calibration Error (ECE), gives a higher weight to low-confidence samples in the minority classes, which leads to inaccurate evaluation of model calibration. Therefore, we also propose Equal Sample Bin Excepted Calibration Error (Esbin-ECE) as a new calibration evaluation metric. Through experiments, we demonstrate that our model yields state-of-the-art in both traditional ECE and Esbin-ECE metrics.