Constraints: Apply penalties on layer parameters

In kerasR: R Interface to the Keras Deep Learning Library

Apply penalties on layer parameters

Description

Regularizers allow to apply penalties on layer parameters or layer activity during optimization. These penalties are incorporated in the loss function that the network optimizes.

Usage

max_norm(max_value = 2, axis = 0)

non_neg()

unit_norm()

Arguments

max_value	maximum value to allow for the value (max_norm only)
axis	axis over which to apply constraint (max_norm only)

Details

The penalties are applied on a per-layer basis. The exact API will depend on the layer, but the layers Dense, Conv1D, Conv2D and Conv3D have a unified API.

Author(s)

Taylor B. Arnold, [e-mail]

References

Chollet, Francois. 2015. Keras: Deep Learning library for Theano and TensorFlow.

Examples

if(keras_available()) {
  X_train <- matrix(rnorm(100 * 10), nrow = 100)
  Y_train <- to_categorical(matrix(sample(0:2, 100, TRUE), ncol = 1), 3)
  
  mod <- Sequential()
  mod$add(Dense(units = 50, input_shape = dim(X_train)[2]))
  mod$add(Activation("relu"))
  mod$add(Dense(units = 3, kernel_constraint = max_norm(),
                bias_constraint = non_neg()))
  mod$add(Dense(units = 3, kernel_constraint = unit_norm()))
  mod$add(Activation("softmax"))
  keras_compile(mod,  loss = 'categorical_crossentropy', optimizer = RMSprop())
  
  keras_fit(mod, X_train, Y_train, batch_size = 32, epochs = 5, verbose = 0)
}

kerasR documentation built on Aug. 17, 2022, 5:06 p.m.