tf.feature_column.embedding_column

TensorFlow 1 version

View source on GitHub

DenseColumn that converts from sparse, categorical input.

tf.feature_column.embedding_column(
    categorical_column, dimension, combiner='mean', initializer=None,
    ckpt_to_load_from=None, tensor_name_in_ckpt=None, max_norm=None, trainable=True,
    use_safe_embedding_lookup=True
)

Use this when your inputs are sparse, but you want to convert them to a dense representation (e.g., to feed to a DNN).

Inputs must be a CategoricalColumn created by any of the categorical_column_* function. Here is an example of using embedding_column with DNNClassifier:

video_id = categorical_column_with_identity(
    key='video_id', num_buckets=1000000, default_value=0)
columns = [embedding_column(video_id, 9),...]

estimator = tf.estimator.DNNClassifier(feature_columns=columns, ...)

label_column = ...
def input_fn():
  features = tf.io.parse_example(
      ..., features=make_parse_example_spec(columns + [label_column]))
  labels = features.pop(label_column.name)
  return features, labels

estimator.train(input_fn=input_fn, steps=100)

Here is an example using embedding_column with model_fn:

def model_fn(features, ...):
  video_id = categorical_column_with_identity(
      key='video_id', num_buckets=1000000, default_value=0)
  columns = [embedding_column(video_id, 9),...]
  dense_tensor = input_layer(features, columns)
  # Form DNN layers, calculate loss, and return EstimatorSpec.
  ...