Neural Clamping: Joint Input Perturbation and Temperature Scaling for Neural Network Calibration [article]

Yung-Chen Tang, Pin-Yu Chen, Tsung-Yi Ho
2022 arXiv   pre-print
Neural network calibration is an essential task in deep learning to ensure consistency between the confidence of model prediction and the true correctness likelihood. In this paper, we propose a new post-processing calibration method called Neural Clamping, which employs a simple joint input-output transformation on a pre-trained classifier via a learnable universal input perturbation and an output temperature scaling parameter. Moreover, we provide theoretical explanations on why Neural
more » ... g is provably better than temperature scaling. Evaluated on CIFAR-100 and ImageNet image recognition datasets and a variety of deep neural network models, our empirical results show that Neural Clamping significantly outperforms state-of-the-art post-processing calibration methods.
arXiv:2209.11604v1 fatcat:67d6ltqirjdetjdzet2uue5yl4