Section outline

  • Loss function


    The loss function is employed to determine if the perceptron is learning. If the model's prediction is accurate, this function will have a low value; otherwise, it will have a high value which means that was unable to correctly classify the data. The loss function is zero when the perceptron has learned all of the inputs correctly. The sum of squared errors as expressed in Eq. 9 is a typical loss function.



    $$ \color{orange} { \sum_{i=0}^{n}(y_{d_i}-y_{e_i})^2 \quad (9) }$$
    \( \color{orange} {\ where: \\ \; \\ y_{d} \ : \ Value \ of \ the \ desired \ output. \\ y_{e} \ : \ Value \ of \ the \ estimated \ output} \)