yadll.objectives

Objectives

mean_squared_error(prediction, target)	Mean Squared Error:
root_mean_squared_error(prediction, target)	Root Mean Squared Error:
mean_absolute_error(prediction, target)	Mean Absolute Error:
binary_hinge_error(prediction, target)	Binary Hinge Error:
categorical_hinge_error(prediction, target)	Categorical Hinge Error:
binary_crossentropy_error(prediction, target)	Binary Cross-entropy Error:
categorical_crossentropy_error(prediction, ...)	Categorical Cross-entropy Error:
kullback_leibler_divergence(prediction, target)	Kullback Leibler Divergence:

Detailed description

yadll.objectives.mean_squared_error(prediction, target)

Mean Squared Error:

\[MSE_i = \frac{1}{n} \sum_{j}{(prediction_{i,j} - target_{i,j})^2}\]

yadll.objectives.root_mean_squared_error(prediction, target)

Root Mean Squared Error:

\[RMSE_i = \sqrt{\frac{1}{n} \sum_{j}{(target_{i,j} - prediction_{i,j})^2}}\]

yadll.objectives.mean_absolute_error(prediction, target)

Mean Absolute Error:

\[MAE_i = \frac{1}{n} \sum_{j}{|target_{i,j} - prediction_{i,j}|}\]

yadll.objectives.binary_hinge_error(prediction, target)

Binary Hinge Error:

\[BHE_i = \frac{1}{n} \sum_{j}{\max(0, 1 - target_{i,j} * prediction_{i,j})}\]

yadll.objectives.categorical_hinge_error(prediction, target)

Categorical Hinge Error:

\[CHE_i = \frac{1}{n} \sum_{j}{\max(1 - target_{i,j} * prediction_{i,j}, 0)}\]

yadll.objectives.binary_crossentropy_error(prediction, target)

Binary Cross-entropy Error:

\[BCE_i = - \frac{1}{n} \sum_{j}{(target_{i,j} * \log(prediction_{i,j}) - (1 - target_{i,j}) * \log(1 - prediction_{i,j}))}\]

yadll.objectives.categorical_crossentropy_error(prediction, target)

Categorical Cross-entropy Error:

\[CCE_i = - \frac{1}{n} \sum_{j}{target_{i,j} * \log(prediction_{i,j})}\]

yadll.objectives.kullback_leibler_divergence(prediction, target)

Kullback Leibler Divergence:

\[KLD_i = \sum_{j}{target_{i,j}*\log(frac{target_{i,j}}{prediction_{i,j}})}\]