error

Value Members

1. object crossentropy extends Error

   

   Annotations

   @lib()

2. object mse extends Error

   

   Mean Squared Error.

   Mean Squared Error.

   This is similar to the SSE with normalization according to dataset-size

   loss: E(w) = 1/2m sum{ (wTx - y)**2 } gradient: dE(w) = 1/m sum{ (wTx - y) *x }

   Annotations

   @lib()

3. object rmse extends Error

   

   Root-Mean-Square Error (RMSE).

   Root-Mean-Square Error (RMSE).

   The root ensures that RMSE is measures on the same scale (and in the same units) as the target variable. This error function is usually used for evaluation and therefore we don't provide the gradient. For trainig the regular MSE works just as well.

   loss: E(w) = sqrt { MSE } gradient: dE(w) = ???

   Annotations

   @lib()

4. object se extends Error

   

   Sum of Squares Error

   Sum of Squares Error

   loss: E(w) = 1/2 sum{ (wTx - y)**2 } gradient: dE(w) = sum{ (wTx - y) *x }

   Annotations

   @lib()