Abstract: Confidence calibration in classification models is a vital technique for accurately estimating the posterior probabilities of predicted results, which is crucial for assessing the likelihood ...