How to regularize models in Keras?

9 months ago

Liam

1 minute

In Keras, regularization can be applied to a model by adding regularization terms within the layers of the model. Regularization terms can be specified in the parameters of each layer, for example:

from keras import regularizers

model = Sequential()
model.add(Dense(64, input_dim=64, kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('relu'))
model.add(Dense(64, kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('relu'))
model.add(Dense(10, kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('softmax'))

In the above example, we added L2 regularization terms with a parameter value of 0.01 to each fully connected layer. Different regularization methods can be chosen as needed, such as L1 regularization, L1L2 regularization, etc. By adding regularization terms, the model will constrain the weights during training to prevent overfitting.