Model Calibration when Subsampling Negatives

Source: Deep Learning on Medium

With class-imbalanced ML problems, it’s often convenient to subsample the negative examples to speed up data pipelines or training jobs…

Continue reading on Medium »