Deep Neural Networks

Create a deep neural network (DNN) estimator.

dnn_regressor(hidden_units, feature_columns, model_dir = NULL,
  label_dimension = 1L, weight_column = NULL, optimizer = "Adagrad",
  activation_fn = "relu", dropout = NULL, input_layer_partitioner = NULL,
  config = NULL)

dnn_classifier(hidden_units, feature_columns, model_dir = NULL,
  n_classes = 2L, weight_column = NULL, label_vocabulary = NULL,
  optimizer = "Adagrad", activation_fn = "relu", dropout = NULL,
  input_layer_partitioner = NULL, config = NULL)

Arguments

hidden_units

An integer vector, indicating the number of hidden units in each layer. All layers are fully connected. For example, c(64, 32) means the first layer has 64 nodes, and the second layer has 32 nodes.

feature_columns

An R list containing all of the feature columns used by the model (typically, generated by feature_columns()).

model_dir

Directory to save the model parameters, graph, and so on. 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 numeric column created by column_numeric() 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 argument. If it is a numeric column, then the raw tensor is fetched by key weight_column$key, then weight_column$normalizer_fn is applied on it to get weight tensor.

optimizer

Either the name of the optimizer to be used when training the model, or a TensorFlow optimizer instance. Defaults to the Adagrad optimizer.

activation_fn

The activation function to apply to each layer. This can either be an actual activation function (e.g. tf$nn$relu), or the name of an activation function (e.g. "relu"). Defaults to the "relu" activation function. See https://www.tensorflow.org/versions/r0.12/api_docs/python/nn/activation_functions_ for documentation related to the set of activation functions available in TensorFlow.

dropout

When not NULL, the probability we will drop out a given coordinate.

input_layer_partitioner

An optional partitioner for the input layer. Defaults to min_max_variable_partitioner with min_slice_size 64 << 20.

config

A run configuration created by run_config(), used to configure the runtime settings.

n_classes

The number of label classes.

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.

See also

Other canned estimators: dnn_linear_combined_estimators, linear_estimators