Skip to content

[source]

Huber Loss#

The pseudo Huber Loss function transitions between L1 and L2 loss at a given pivot point (defined by delta) such that the function becomes more quadratic as the loss decreases. The combination of L1 and L2 losses make Huber more robust to outliers while maintaining smoothness near the minimum.

\[ L_{\delta}= \left\{\begin{matrix} \frac{1}{2}(y - \hat{y})^{2} & if \left | (y - \hat{y}) \right | < \delta\\ \delta ((y - \hat{y}) - \frac1 2 \delta) & otherwise \end{matrix}\right. \]

Parameters#

# Name Default Type Description
1 delta 1.0 float The pivot point i.e the point where numbers larger will be evaluated with an L1 loss while number smaller will be evaluated with an L2 loss.

Example#

use Rubix\ML\NeuralNet\CostFunctions\HuberLoss;

$costFunction = new HuberLoss(0.5);