How to create a custom cost function to evaluate keras model?

Recipe Objective

How to create a custom cost function to evaluate keras model

The formula for every loss function is predefined, but if we want to create a loss function(cost function) specifically for our model then we can create. So we can define our own loss function and call it a custom cost function.

Step 1- Importing Libraries

import keras as k from keras.models import Sequential from keras.layers import Dense import numpy as np

Step 2- Defining two sample arrays.

We will define two sample arrays as predicted and actual to calculate the loss. y_pred=np.array([2,3,5,7,9]) y_actual=np.array([4,2,8,5,2])

Step 3- Define your new custom loss function.

we are considering the formula for MSE here.

def custom_loss(y_true,y_pred): return K.mean(K.square(y_pred - y_actual) + K.square(layer), axis=-1)

Step 4- Creating the Neural Network Model.

We will pass our own custom cost function to the model.

# create model model = Sequential() model.add(Dense(12, input_dim=8, activation='relu')) model.add(Dense(8, activation='relu')) model.add(Dense(1, activation='sigmoid')) # Compile model model.compile(loss='custom_loss', optimizer='adam', metrics=['accuracy'])

Step 5- Create the model summary.

model.summary

>