tfl.lattice_layer.TorsionRegularizer

View source on GitHub

Class TorsionRegularizer

Torsion regularizer for tfl.layers.Lattice layer.

Lattice torsion regularizer penalizes how much the lattice function twists from side-to-side (see publication).

Consider a 3 x 2 lattice with weights w:

w[3]-----w[4]-----w[5]
  |        |        |
  |        |        |
w[0]-----w[1]-----w[2]

In this case, the torsion regularizer is defined as:

l1 * (|w[4] + w[0] - w[3] - w[1]| + |w[5] + w[1] - w[4] - w[2]|) +
l2 * ((w[4] + w[0] - w[3] - w[1])^2 + (w[5] + w[1] - w[4] - w[2])^2)

Attributes:

  • All __init__ arguments.

__init__

View source

__init__(
    lattice_sizes,
    l1=0.0,
    l2=0.0
)

Initializes an instance of TorsionRegularizer.

Args:

  • lattice_sizes: Lattice sizes of tfl.layers.Lattice to regularize.
  • l1: l1 regularization amount. Either single float or list or tuple of floats to specify different regularization amount per dimension. The amount of regularization for the interaction term between two dimensions is the product of the corresponding per dimension amounts.
  • l2: l2 regularization amount. Either single float or list or tuple of floats to specify different regularization amount per dimension. The amount of regularization for the interaction term between two dimensions is the product of the corresponding per dimension amounts.

Methods

__call__

View source

__call__(x)

Returns regularization loss for x.

from_config

@classmethod
from_config(
    cls,
    config
)

Creates a regularizer from its config.

This method is the reverse of get_config, capable of instantiating the same regularizer from the config dictionary.

This method is used by Keras model_to_estimator, saving and loading models to HDF5 formats, Keras model cloning, some visualization utilities, and exporting models to and from JSON.

Arguments:

  • config: A Python dictionary, typically the output of get_config.

Returns:

A regularizer instance.

get_config

View source

get_config()

Standard Keras config for serialization.