Improving Calibration in Mixup-trained Deep Neural Networks through Confidence-Based Loss Functions

Open in new window