Registro de evidencia del método

Model Calibration

Model calibration is a post-hoc technique that adjusts the probability outputs of a trained classifier so that predicted confidence scores match empirical outcome frequencies. A classifier is said to be perfectly calibrated if, among all predictions made with confidence p, exactly a fraction p of them are correct. Systematic miscalibration of modern deep neural networks was rigorously documented by Guo et al. (2017), who showed that networks trained with standard cross-entropy loss tend to be overconfident, and proposed temperature scaling as a simple, effective remedy.

Sources recorded, not reviewed

Registro de origen

Citas copiadas textualmente del registro de origen del método. No se infiere ninguna verificación a nivel de afirmación de ellas.

Probability Calibration of Classifiers

Registro del método taxonómico · ml-model / machine-learning

Guo, C., Pleiss, G., Sun, Y., & Weinberger, K. Q. (2017). On calibration of modern neural networks. International Conference on Machine Learning, 1321–1330. · URL

Abrir método completo

Afirmaciones curadas

Afirmaciones persistidas en el libro mayor de evidencia, cada una con su propia evaluación.

Aún no hay afirmaciones curadas

Esta vista no inventa una evaluación de afirmación si el libro mayor no tiene ninguna.

Métodos relacionados

Generado a partir del grafo de métodos y mostrado como relaciones sugeridas por la máquina; no se infiere ninguna afirmación de evidencia.

Taxonomic bucketConformal Predictionmachine-suggested · Relational suggestion, not evidence.See alsoLogistic Regressionmachine-suggested · Relational suggestion, not evidence.See alsoUncertainty Quantificationmachine-suggested · Relational suggestion, not evidence.