tf.keras.regularizers.l1_l2

TensorFlow 1 version View source on GitHub

Create a regularizer that applies both L1 and L2 penalties.

tf.keras.regularizers.l1_l2(
    l1=0.01, l2=0.01
)

The L1 regularization penalty is computed as:

\(\ell_1\,\,penalty =\ell_1\sum_{i=0}^n|x_i|\)

The L2 regularization penalty is computed as:

\(\ell_2\,\,penalty =\ell_2\sum_{i=0}^nx_i^2\)

Arguments:

  • l1: Float; L1 regularization factor.
  • l2: Float; L2 regularization factor.

Returns:

An L1L2 Regularizer with the given regularization factors.