estimator/canned/dnn.py

# Copyright 2017 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Deep Neural Network estimators."""

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import six

from tensorflow.python.estimator import estimator
from tensorflow.python.estimator import model_fn
from tensorflow.python.estimator.canned import head as head_lib
from tensorflow.python.estimator.canned import optimizers
from tensorflow.python.feature_column import feature_column as feature_column_lib
from tensorflow.python.layers import core as core_layers
from tensorflow.python.ops import init_ops
from tensorflow.python.ops import nn
from tensorflow.python.ops import partitioned_variables
from tensorflow.python.ops import variable_scope
from tensorflow.python.ops.losses import losses
from tensorflow.python.summary import summary
from tensorflow.python.training import training_util
from tensorflow.python.util.tf_export import tf_export

# The default learning rate of 0.05 is a historical artifact of the initial
# implementation, but seems a reasonable choice.
_LEARNING_RATE = 0.05


def _add_hidden_layer_summary(value, tag):
  summary.scalar('%s/fraction_of_zero_values' % tag, nn.zero_fraction(value))
  summary.histogram('%s/activation' % tag, value)


def _dnn_logit_fn_builder(units, hidden_units, feature_columns, activation_fn,
                          dropout, input_layer_partitioner):
  """Function builder for a dnn logit_fn.

  Args:
    units: An int indicating the dimension of the logit layer.  In the
      MultiHead case, this should be the sum of all component Heads' logit
      dimensions.
    hidden_units: Iterable of integer number of hidden units per layer.
    feature_columns: Iterable of `feature_column._FeatureColumn` model inputs.
    activation_fn: Activation function applied to each layer.
    dropout: When not `None`, the probability we will drop out a given
      coordinate.
    input_layer_partitioner: Partitioner for input layer.

  Returns:
    A logit_fn (see below).

  Raises:
    ValueError: If units is not an int.
  """
  if not isinstance(units, int):
    raise ValueError('units must be an int.  Given type: {}'.format(
        type(units)))

  def dnn_logit_fn(features, mode):
    """Deep Neural Network logit_fn.

    Args:
      features: This is the first item returned from the `input_fn`
                passed to `train`, `evaluate`, and `predict`. This should be a
                single `Tensor` or `dict` of same.
      mode: Optional. Specifies if this training, evaluation or prediction. See
            `ModeKeys`.

    Returns:
      A `Tensor` representing the logits, or a list of `Tensor`'s representing
      multiple logits in the MultiHead case.
    """
    with variable_scope.variable_scope(
        'input_from_feature_columns',
        values=tuple(six.itervalues(features)),
        partitioner=input_layer_partitioner):
      net = feature_column_lib.input_layer(
          features=features, feature_columns=feature_columns)
    for layer_id, num_hidden_units in enumerate(hidden_units):
      with variable_scope.variable_scope(
          'hiddenlayer_%d' % layer_id, values=(net,)) as hidden_layer_scope:
        net = core_layers.dense(
            net,
            units=num_hidden_units,
            activation=activation_fn,
            kernel_initializer=init_ops.glorot_uniform_initializer(),
            name=hidden_layer_scope)
        if dropout is not None and mode == model_fn.ModeKeys.TRAIN:
          net = core_layers.dropout(net, rate=dropout, training=True)
      _add_hidden_layer_summary(net, hidden_layer_scope.name)

    with variable_scope.variable_scope('logits', values=(net,)) as logits_scope:
      logits = core_layers.dense(
          net,
          units=units,
          activation=None,
          kernel_initializer=init_ops.glorot_uniform_initializer(),
          name=logits_scope)
    _add_hidden_layer_summary(logits, logits_scope.name)

    return logits

  return dnn_logit_fn


def _dnn_model_fn(features,
                  labels,
                  mode,
                  head,
                  hidden_units,
                  feature_columns,
                  optimizer='Adagrad',
                  activation_fn=nn.relu,
                  dropout=None,
                  input_layer_partitioner=None,
                  config=None):
  """Deep Neural Net model_fn.

  Args:
    features: dict of `Tensor`.
    labels: `Tensor` of shape [batch_size, 1] or [batch_size] labels of
      dtype `int32` or `int64` in the range `[0, n_classes)`.
    mode: Defines whether this is training, evaluation or prediction.
      See `ModeKeys`.
    head: A `head_lib._Head` instance.
    hidden_units: Iterable of integer number of hidden units per layer.
    feature_columns: Iterable of `feature_column._FeatureColumn` model inputs.
    optimizer: String, `tf.Optimizer` object, or callable that creates the
      optimizer to use for training. If not specified, will use the Adagrad
      optimizer with a default learning rate of 0.05.
    activation_fn: Activation function applied to each layer.
    dropout: When not `None`, the probability we will drop out a given
      coordinate.
    input_layer_partitioner: Partitioner for input layer. Defaults
      to `min_max_variable_partitioner` with `min_slice_size` 64 << 20.
    config: `RunConfig` object to configure the runtime settings.

  Returns:
    An `EstimatorSpec` instance.

  Raises:
    ValueError: If features has the wrong type.
  """
  if not isinstance(features, dict):
    raise ValueError('features should be a dictionary of `Tensor`s. '
                     'Given type: {}'.format(type(features)))

  optimizer = optimizers.get_optimizer_instance(
      optimizer, learning_rate=_LEARNING_RATE)
  num_ps_replicas = config.num_ps_replicas if config else 0

  partitioner = partitioned_variables.min_max_variable_partitioner(
      max_partitions=num_ps_replicas)
  with variable_scope.variable_scope(
      'dnn',
      values=tuple(six.itervalues(features)),
      partitioner=partitioner):
    input_layer_partitioner = input_layer_partitioner or (
        partitioned_variables.min_max_variable_partitioner(
            max_partitions=num_ps_replicas,
            min_slice_size=64 << 20))

    logit_fn = _dnn_logit_fn_builder(
        units=head.logits_dimension,
        hidden_units=hidden_units,
        feature_columns=feature_columns,
        activation_fn=activation_fn,
        dropout=dropout,
        input_layer_partitioner=input_layer_partitioner)
    logits = logit_fn(features=features, mode=mode)

    def _train_op_fn(loss):
      """Returns the op to optimize the loss."""
      return optimizer.minimize(
          loss,
          global_step=training_util.get_global_step())

    return head.create_estimator_spec(
        features=features,
        mode=mode,
        labels=labels,
        train_op_fn=_train_op_fn,
        logits=logits)


@tf_export('estimator.DNNClassifier')
class DNNClassifier(estimator.Estimator):
  """A classifier for TensorFlow DNN models.

  Example:

  ```python
  categorical_feature_a = categorical_column_with_hash_bucket(...)
  categorical_feature_b = categorical_column_with_hash_bucket(...)

  categorical_feature_a_emb = embedding_column(
      categorical_column=categorical_feature_a, ...)
  categorical_feature_b_emb = embedding_column(
      categorical_column=categorical_feature_b, ...)

  estimator = DNNClassifier(
      feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
      hidden_units=[1024, 512, 256])

  # Or estimator using the ProximalAdagradOptimizer optimizer with
  # regularization.
  estimator = DNNClassifier(
      feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
      hidden_units=[1024, 512, 256],
      optimizer=tf.train.ProximalAdagradOptimizer(
        learning_rate=0.1,
        l1_regularization_strength=0.001
      ))

  # Or estimator with warm-starting from a previous checkpoint.
  estimator = DNNClassifier(
      feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
      hidden_units=[1024, 512, 256],
      warm_start_from="/path/to/checkpoint/dir")

  # Input builders
  def input_fn_train: # returns x, y
    pass
  estimator.train(input_fn=input_fn_train, steps=100)

  def input_fn_eval: # returns x, y
    pass
  metrics = estimator.evaluate(input_fn=input_fn_eval, steps=10)
  def input_fn_predict: # returns x, None
    pass
  predictions = estimator.predict(input_fn=input_fn_predict)
  ```

  Input of `train` and `evaluate` should have following features,
  otherwise there will be a `KeyError`:

  * if `weight_column` is not `None`, a feature with
    `key=weight_column` whose value is a `Tensor`.
  * for each `column` in `feature_columns`:
    - if `column` is a `_CategoricalColumn`, a feature with `key=column.name`
      whose `value` is a `SparseTensor`.
    - if `column` is a `_WeightedCategoricalColumn`, two features: the first
      with `key` the id column name, the second with `key` the weight column
      name. Both features' `value` must be a `SparseTensor`.
    - if `column` is a `_DenseColumn`, a feature with `key=column.name`
      whose `value` is a `Tensor`.

  Loss is calculated by using softmax cross entropy.

  @compatibility(eager)
  Estimators are not compatible with eager execution.
  @end_compatibility
  """

  def __init__(
      self,
      hidden_units,
      feature_columns,
      model_dir=None,
      n_classes=2,
      weight_column=None,
      label_vocabulary=None,
      optimizer='Adagrad',
      activation_fn=nn.relu,
      dropout=None,
      input_layer_partitioner=None,
      config=None,
      warm_start_from=None,
      loss_reduction=losses.Reduction.SUM,
  ):
    """Initializes a `DNNClassifier` instance.

    Args:
      hidden_units: Iterable of number hidden units per layer. All layers are
        fully connected. Ex. `[64, 32]` means first layer has 64 nodes and
        second one has 32.
      feature_columns: An iterable containing all the feature columns used by
        the model. All items in the set should be instances of classes derived
        from `_FeatureColumn`.
      model_dir: Directory to save model parameters, graph and etc. This can
        also be used to load checkpoints from the directory into a estimator to
        continue training a previously saved model.
      n_classes: Number of label classes. Defaults to 2, namely binary
        classification. Must be > 1.
      weight_column: A string or a `_NumericColumn` created by
        `tf.feature_column.numeric_column` defining feature column representing
        weights. It is used to down weight or boost examples during training. It
        will be multiplied by the loss of the example. If it is a string, it is
        used as a key to fetch weight tensor from the `features`. If it is a
        `_NumericColumn`, raw tensor is fetched by key `weight_column.key`,
        then weight_column.normalizer_fn is applied on it to get weight tensor.
      label_vocabulary: A list of strings represents possible label values. If
        given, labels must be string type and have any value in
        `label_vocabulary`. If it is not given, that means labels are
        already encoded as integer or float within [0, 1] for `n_classes=2` and
        encoded as integer values in {0, 1,..., n_classes-1} for `n_classes`>2 .
        Also there will be errors if vocabulary is not provided and labels are
        string.
      optimizer: An instance of `tf.Optimizer` used to train the model. Defaults
        to Adagrad optimizer.
      activation_fn: Activation function applied to each layer. If `None`, will
        use `tf.nn.relu`.
      dropout: When not `None`, the probability we will drop out a given
        coordinate.
      input_layer_partitioner: Optional. Partitioner for input layer. Defaults
        to `min_max_variable_partitioner` with `min_slice_size` 64 << 20.
      config: `RunConfig` object to configure the runtime settings.
      warm_start_from: A string filepath to a checkpoint to warm-start from, or
        a `WarmStartSettings` object to fully configure warm-starting.  If the
        string filepath is provided instead of a `WarmStartSettings`, then all
        weights are warm-started, and it is assumed that vocabularies and Tensor
        names are unchanged.
      loss_reduction: One of `tf.losses.Reduction` except `NONE`. Describes how
        to reduce training loss over batch. Defaults to `SUM`.
    """
    if n_classes == 2:
      head = head_lib._binary_logistic_head_with_sigmoid_cross_entropy_loss(  # pylint: disable=protected-access
          weight_column=weight_column,
          label_vocabulary=label_vocabulary,
          loss_reduction=loss_reduction)
    else:
      head = head_lib._multi_class_head_with_softmax_cross_entropy_loss(  # pylint: disable=protected-access
          n_classes, weight_column=weight_column,
          label_vocabulary=label_vocabulary,
          loss_reduction=loss_reduction)

    def _model_fn(features, labels, mode, config):
      """Call the defined shared _dnn_model_fn."""
      return _dnn_model_fn(
          features=features,
          labels=labels,
          mode=mode,
          head=head,
          hidden_units=hidden_units,
          feature_columns=tuple(feature_columns or []),
          optimizer=optimizer,
          activation_fn=activation_fn,
          dropout=dropout,
          input_layer_partitioner=input_layer_partitioner,
          config=config)

    super(DNNClassifier, self).__init__(
        model_fn=_model_fn, model_dir=model_dir, config=config,
        warm_start_from=warm_start_from)


@tf_export('estimator.DNNRegressor')
class DNNRegressor(estimator.Estimator):
  """A regressor for TensorFlow DNN models.

  Example:

  ```python
  categorical_feature_a = categorical_column_with_hash_bucket(...)
  categorical_feature_b = categorical_column_with_hash_bucket(...)

  categorical_feature_a_emb = embedding_column(
      categorical_column=categorical_feature_a, ...)
  categorical_feature_b_emb = embedding_column(
      categorical_column=categorical_feature_b, ...)

  estimator = DNNRegressor(
      feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
      hidden_units=[1024, 512, 256])

  # Or estimator using the ProximalAdagradOptimizer optimizer with
  # regularization.
  estimator = DNNRegressor(
      feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
      hidden_units=[1024, 512, 256],
      optimizer=tf.train.ProximalAdagradOptimizer(
        learning_rate=0.1,
        l1_regularization_strength=0.001
      ))

  # Or estimator with warm-starting from a previous checkpoint.
  estimator = DNNRegressor(
      feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
      hidden_units=[1024, 512, 256],
      warm_start_from="/path/to/checkpoint/dir")

  # Input builders
  def input_fn_train: # returns x, y
    pass
  estimator.train(input_fn=input_fn_train, steps=100)

  def input_fn_eval: # returns x, y
    pass
  metrics = estimator.evaluate(input_fn=input_fn_eval, steps=10)
  def input_fn_predict: # returns x, None
    pass
  predictions = estimator.predict(input_fn=input_fn_predict)
  ```

  Input of `train` and `evaluate` should have following features,
  otherwise there will be a `KeyError`:

  * if `weight_column` is not `None`, a feature with
    `key=weight_column` whose value is a `Tensor`.
  * for each `column` in `feature_columns`:
    - if `column` is a `_CategoricalColumn`, a feature with `key=column.name`
      whose `value` is a `SparseTensor`.
    - if `column` is a `_WeightedCategoricalColumn`, two features: the first
      with `key` the id column name, the second with `key` the weight column
      name. Both features' `value` must be a `SparseTensor`.
    - if `column` is a `_DenseColumn`, a feature with `key=column.name`
      whose `value` is a `Tensor`.

  Loss is calculated by using mean squared error.

  @compatibility(eager)
  Estimators are not compatible with eager execution.
  @end_compatibility
  """

  def __init__(
      self,
      hidden_units,
      feature_columns,
      model_dir=None,
      label_dimension=1,
      weight_column=None,
      optimizer='Adagrad',
      activation_fn=nn.relu,
      dropout=None,
      input_layer_partitioner=None,
      config=None,
      warm_start_from=None,
      loss_reduction=losses.Reduction.SUM,
  ):
    """Initializes a `DNNRegressor` instance.

    Args:
      hidden_units: Iterable of number hidden units per layer. All layers are
        fully connected. Ex. `[64, 32]` means first layer has 64 nodes and
        second one has 32.
      feature_columns: An iterable containing all the feature columns used by
        the model. All items in the set should be instances of classes derived
        from `_FeatureColumn`.
      model_dir: Directory to save model parameters, graph and etc. This can
        also be used to load checkpoints from the directory into a estimator to
        continue training a previously saved model.
      label_dimension: Number of regression targets per example. This is the
        size of the last dimension of the labels and logits `Tensor` objects
        (typically, these have shape `[batch_size, label_dimension]`).
      weight_column: A string or a `_NumericColumn` created by
        `tf.feature_column.numeric_column` defining feature column representing
        weights. It is used to down weight or boost examples during training. It
        will be multiplied by the loss of the example. If it is a string, it is
        used as a key to fetch weight tensor from the `features`. If it is a
        `_NumericColumn`, raw tensor is fetched by key `weight_column.key`,
        then weight_column.normalizer_fn is applied on it to get weight tensor.
      optimizer: An instance of `tf.Optimizer` used to train the model. Defaults
        to Adagrad optimizer.
      activation_fn: Activation function applied to each layer. If `None`, will
        use `tf.nn.relu`.
      dropout: When not `None`, the probability we will drop out a given
        coordinate.
      input_layer_partitioner: Optional. Partitioner for input layer. Defaults
        to `min_max_variable_partitioner` with `min_slice_size` 64 << 20.
      config: `RunConfig` object to configure the runtime settings.
      warm_start_from: A string filepath to a checkpoint to warm-start from, or
        a `WarmStartSettings` object to fully configure warm-starting.  If the
        string filepath is provided instead of a `WarmStartSettings`, then all
        weights are warm-started, and it is assumed that vocabularies and Tensor
        names are unchanged.
      loss_reduction: One of `tf.losses.Reduction` except `NONE`. Describes how
        to reduce training loss over batch. Defaults to `SUM`.
    """

    def _model_fn(features, labels, mode, config):
      """Call the defined shared _dnn_model_fn."""
      return _dnn_model_fn(
          features=features,
          labels=labels,
          mode=mode,
          head=head_lib.  # pylint: disable=protected-access
          _regression_head_with_mean_squared_error_loss(
              label_dimension=label_dimension, weight_column=weight_column,
              loss_reduction=loss_reduction),
          hidden_units=hidden_units,
          feature_columns=tuple(feature_columns or []),
          optimizer=optimizer,
          activation_fn=activation_fn,
          dropout=dropout,
          input_layer_partitioner=input_layer_partitioner,
          config=config)

    super(DNNRegressor, self).__init__(
        model_fn=_model_fn, model_dir=model_dir, config=config,
        warm_start_from=warm_start_from)