KerasRS
API documentation Embedding Layers Feature Interaction Layers Retrieval Layers Losses Metrics Examples
► KerasRS / API documentation / Feature Interaction Layers / DotInteraction layer

DotInteraction layer

[source]

DotInteraction class

keras_rs.layers.DotInteraction(
    self_interaction: bool = False, skip_gather: bool = False, **kwargs: Any
)

Dot interaction layer present in DLRM.

This layer computes distinct dot product ("feature interactions") for every pair of features. If self_interaction is True, we calculate dot products of the form dot(e_i, e_j) for i <= j, and dot(e_i, e_j) for i < j, otherwise. e_i denotes representation of feature i. The layer can be used to build the DLRM model.

Arguments

  • self_interaction: bool. Indicates whether features should "self-interact". If it is True, then the diagonal entries of the interaction matrix are also taken.
  • skip_gather: bool. If it's set, then the upper triangular part of the interaction matrix is set to 0. The output will be of shape [batch_size, num_features * num_features] of which half of the entries will be zeros. Otherwise, the output will be only the lower triangular part of the interaction matrix. The latter saves space but is much slower.
  • **kwargs: Args to pass to the base class.

Example

# 1. Simple forward pass
batch_size = 2
embedding_dim = 32
feature1 = np.random.randn(batch_size, embedding_dim)
feature2 = np.random.randn(batch_size, embedding_dim)
feature3 = np.random.randn(batch_size, embedding_dim)
feature_interactions = keras_rs.layers.DotInteraction()(
    [feature1, feature2, feature3]
)

# 2. After embedding layer in a model
vocabulary_size = 32
embedding_dim = 6

# Create a simple model containing the layer.
feature_input_1 = keras.Input(shape=(), name='indices_1', dtype="int32")
feature_input_2 = keras.Input(shape=(), name='indices_2', dtype="int32")
feature_input_3 = keras.Input(shape=(), name='indices_3', dtype="int32")
x1 = keras.layers.Embedding(
    input_dim=vocabulary_size,
    output_dim=embedding_dim
)(feature_input_1)
x2 = keras.layers.Embedding(
    input_dim=vocabulary_size,
    output_dim=embedding_dim
)(feature_input_2)
x3 = keras.layers.Embedding(
    input_dim=vocabulary_size,
    output_dim=embedding_dim
)(feature_input_3)
feature_interactions = keras_rs.layers.DotInteraction()([x1, x2, x3])
output = keras.layers.Dense(units=10)(x2)
model = keras.Model(
    [feature_input_1, feature_input_2, feature_input_3], output
)

# Call the model on the inputs.
batch_size = 2
f1 = np.random.randint(0, vocabulary_size, size=(batch_size,))
f2 = np.random.randint(0, vocabulary_size, size=(batch_size,))
f3 = np.random.randint(0, vocabulary_size, size=(batch_size,))
outputs = model([f1, f2, f3])

References

[source]

call method

DotInteraction.call(inputs: list[typing.Any])

Forward pass of the dot interaction layer.

Arguments

  • inputs: list. Every element in the list represents a feature tensor of shape [batch_size, feature_dim]. All tensors in the list must have the same shape.

Returns

Tensor representing feature interactions. The shape of the tensor is [batch_size, k] where k is num_features * num_features if skip_gather is True. Otherwise, k is num_features * (num_features + 1) / 2 if self_interaction is True, and num_features * (num_features - 1) / 2 if not.

DotInteraction layer
DotInteraction class
call method