Distribution-Free Calibration Guarantees for Histogram Binning
without Sample Splitting
Chirag Gupta 1 Aaditya Ramdas 1
Abstract the goal is to recalibrate existing (‘pre-learnt’) classifiers
that are powerful and (statistically) efficient for classifi-
We prove calibration guarantees for the popu-
cation accuracy, but do not satisfy calibration properties
...


雷达卡




京公网安备 11010802022788号







