Glosarium AI
Kamus lengkap Kecerdasan Buatan
Calibration Plot
Visualization graph that compares predicted probabilities to observed empirical frequencies to visually evaluate the calibration of a classification model.
Proper Scoring Rule
Evaluation function that incentivizes honest predictions by rewarding only when predicted probabilities match the true probability distributions.
Sharpness
Property of probabilistic predictions that measures their concentration or dispersion, independently of their calibration, evaluated by the variance of predicted probabilities.
Multi-class Brier Score
Extension of the Brier Score to multi-class classification problems that calculates the sum of squared differences between predicted probabilities and one-hot target vectors.
Calibration Refinement
Iterative process of calibration improvement that combines recalibration techniques with evaluation metrics to optimize calibration parameters.