arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2109.05263cs.CV
收藏

基于类别分布的长尾视觉识别校正

Class-Distribution-Aware Calibration for Long-Tailed Visual Recognition

Imperial College London(伦敦帝国理工学院);National University of Singapore(新加坡国立大学)

Mobarakol Islam, Lalithkumar Seenivasan, Hongliang Ren, Ben Glocker

摘要:

尽管深度神经网络的准确性令人印象深刻,但它经常被错误校准,并且往往会做出过于自信的预测。最近的技术,如温度缩放(TS)和标签平滑(LS)显示出有效性,在获得一个良好的校准模型,通过平滑的logits和硬标签与标量因子,分别。然而,使用统一的TS或LS因子对于校准在长尾数据集上训练的模型可能不是最佳的,在长尾数据集上,模型对高频类别产生过度自信的概率。在这项研究中,我们提出了类分布感知TS(CDA-TS)和LS(CDA-LS),将类频率信息纳入模型校准的长尾分布的背景下。在CDA-TS中,标量温度值被用类别频率编码的CDA温度向量替换以补偿过度置信。类似地,CDA-LS使用向量平滑因子,并根据其对应的类分布对硬标签进行平滑。我们还将CDA最佳温度矢量与蒸馏损失相结合,减少了自蒸馏(SD)中的误校准。我们的经验表明,类分布感知TS和LS可以适应不平衡的数据分布,从而在校准误差和预测精度方面都具有优异的性能。我们还观察到,SD与一个非常不平衡的数据集是不太有效的校准性能。代码可在https://github.com/mobarakol/Class-Distribution-Aware-TS-LS上找到。

英文摘要:

Despite impressive accuracy, deep neural networks are often miscalibrated and tend to overly confident predictions. Recent techniques like temperature scaling (TS) and label smoothing (LS) show effectiveness in obtaining a well-calibrated model by smoothing logits and hard labels with scalar factors, respectively. However, the use of uniform TS or LS factor may not be optimal for calibrating models trained on a long-tailed dataset where the model produces overly confident probabilities for high-frequency classes. In this study, we propose class-distribution-aware TS (CDA-TS) and LS (CDA-LS) by incorporating class frequency information in model calibration in the context of long-tailed distribution. In CDA-TS, the scalar temperature value is replaced with the CDA temperature vector encoded with class frequency to compensate for the over-confidence. Similarly, CDA-LS uses a vector smoothing factor and flattens the hard labels according to their corresponding class distribution. We also integrate CDA optimal temperature vector with distillation loss, which reduces miscalibration in self-distillation (SD). We empirically show that class-distribution-aware TS and LS can accommodate the imbalanced data distribution yielding superior performance in both calibration error and predictive accuracy. We also observe that SD with an extremely imbalanced dataset is less effective in terms of calibration performance. Code is available in https://github.com/mobarakol/Class-Distribution-Aware-TS-LS.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑