Perte de Huber

Perte de Huber

Perte de Huber is a popular fonction de perte used in regression problems, particularly in apprentissage automatique and statistics. It combines the advantages of two other des fonctions de perte: erreur quadratique moyenne (MSE) and erreur absolue moyenne (MAE). Unlike MSE, which can be heavily influenced by outliers due to the squaring of errors, Huber Loss is designed to be robust against such anomalies.

La perte de Huber est définie par un paramètre appelé seuil (souvent noté δ), which determines the point at which the loss function transitions from quadratic to linear. For residuals (the differences between actual and predicted values) that are less than δ en valeur absolue, la perte de Huber se comporte comme la MSE, en utilisant la formule :

Perte de Huber = 0,5 * (résidu)^2

Pour les résidus qui dépassent δ, the loss is calculated using the absolute error formula, which is less sensitive to large errors:

Huber Loss = δ * (|residual| – 0.5 * δ)

Cette combinaison permet à la Perte de Huber de fournir un gradient fluide pour optimization while limiting the influence of outliers. When selecting δ, it is important to consider the scale of the data and the specific characteristics of the dataset.

Huber Loss is particularly useful in scenarios where a dataset contains outliers that could skew the results if MSE were used exclusively. It strikes a balance between maintaining sensitivity to small errors and robustness against large deviations, making it a versatile choice for many regression applications.