fusionlab.nn.components.StaticEnrichmentLayer

class fusionlab.nn.components.StaticEnrichmentLayer

Bases: Layer, NNLearner

Static Enrichment Layer for combining static and temporal features [1].

This layer enriches temporal features with static context, enabling the model to modulate temporal dynamics based on static information. It concatenates a tiled static context vector to temporal features and processes them through a GatedResidualNetwork, yielding an enriched feature map that combines both static and temporal information.

\[\mathbf{Z} = \text{GRN}\big([\mathbf{C}, \mathbf{X}]\big)\]

where \(\mathbf{C}\) is a static context vector tiled over the time dimension, and \(\mathbf{X}\) are the temporal features.

Parameters:

• units (int) – Number of hidden units within the internally used GatedResidualNetwork.

• activation (str, optional) – Activation function used in the GRN. Must be one of {‘elu’, ‘relu’, ‘tanh’, ‘sigmoid’, ‘linear’}. Defaults to 'elu'.

• use_batch_norm (bool, optional) – Whether to apply batch normalization within the GRN. Defaults to False.

• **kwargs – Additional arguments passed to the parent Keras Layer.

Notes

This layer performs the following: 1. Expand static context from shape

\((B, U)\) to \((B, T, U)\).

2. Concatenate with temporal features \((B, T, D)\) along the last dimension.

3. Pass the combined tensor through a GatedResidualNetwork.

call(`static_context_vector`, `temporal_features`,

training=False)

Forward pass of the static enrichment layer.

get_config()

Returns the configuration dictionary for serialization.

from_config(`config`)

Instantiates the layer from a configuration dictionary.

Examples

>>> from fusionlab.nn.components import StaticEnrichmentLayer
>>> import tensorflow as tf
>>> # Define static context of shape (batch_size, units)
... # and temporal features of shape
... # (batch_size, time_steps, units)
>>> static_context_vector = tf.random.normal((32, 64))
>>> temporal_features = tf.random.normal((32, 10, 64))
>>> # Instantiate the static enrichment layer
>>> sel = StaticEnrichmentLayer(
...     units=64,
...     activation='relu',
...     use_batch_norm=True
... )
>>> # Forward pass
>>> outputs = sel(
...     static_context_vector,
...     temporal_features,
...     training=True
... )

See also

GatedResidualNetwork

Used within the static enrichment layer to combine static and temporal features.

TemporalFusionTransformer

Incorporates the static enrichment mechanism.

References

[1]

Lim, B., & Zohren, S. (2021). “Time-series forecasting with deep learning: a survey.” Philosophical Transactions of the Royal Society A, 379(2194), 20200209.

__init__(units, activation='elu', use_batch_norm=False, **kwargs)

Initialize the StaticEnrichmentLayer.

Parameters:

• units (int) – Number of hidden units in the internal GatedResidualNetwork.

• activation (str, optional) – Activation function for the GRN. Defaults to 'elu'.

• use_batch_norm (bool, optional) – Whether to apply batch normalization in the GRN. Defaults to False.

• **kwargs – Additional arguments passed to the parent Keras Layer.

Methods

__init__(units[, activation, use_batch_norm])	Initialize the StaticEnrichmentLayer.
add_loss(loss)	Can be called inside of the call() method to add a scalar loss.
add_metric(*args, **kwargs)
add_variable(shape, initializer[, dtype, ...])	Add a weight variable to the layer.
add_weight([shape, initializer, dtype, ...])	Add a weight variable to the layer.
build(input_shape)
build_from_config(config)	Builds the layer's states with the supplied config dict.
call(temporal_features, context_vector[, ...])	Forward pass of the static enrichment layer.
compute_mask(inputs, previous_mask)
compute_output_shape(*args, **kwargs)
compute_output_spec(*args, **kwargs)
count_params()	Count the total number of scalars composing the weights.
from_config(config)	Create a new instance from a config dictionary.
get_build_config()	Returns a dictionary with the layer's input shape.
get_config()	Return the layer configuration for serialization.
get_params([deep])	Get the parameters for this learner.
get_weights()	Return the values of layer.weights as a list of NumPy arrays.
help(**kwargs)
load(file_path[, format])	Load the learner's state from a specified file in the desired format.
load_own_variables(store)	Loads the state of the layer.
quantize(mode[, type_check, config])
quantized_build(input_shape, mode)
quantized_call(*args, **kwargs)
rematerialized_call(layer_call, *args, **kwargs)	Enable rematerialization dynamically for layer's call method.
save([file_path, format, overwrite, ...])	Save the learner's state to a specified file in the desired format.
save_own_variables(store)	Saves the state of the layer.
set_params(**params)	Set the parameters of this learner.
set_weights(weights)	Sets the values of layer.weights from a list of NumPy arrays.
stateless_call(trainable_variables, ...[, ...])	Call the layer without any side effects.
summary()	Provide a summary of the learner's parameters.
symbolic_call(*args, **kwargs)

Attributes

compute_dtype	The dtype of the computations performed by the layer.
dtype	Alias of layer.variable_dtype.
dtype_policy
input	Retrieves the input tensor(s) of a symbolic operation.
input_dtype	The dtype layer inputs should be converted to.
input_spec
losses	List of scalar losses from add_loss, regularizers and sublayers.
metrics	List of all metrics.
metrics_variables	List of all metric variables.
my_params
non_trainable_variables	List of all non-trainable layer state.
non_trainable_weights	List of all non-trainable weight variables of the layer.
output	Retrieves the output tensor(s) of a layer.
path	The path of the layer.
quantization_mode	The quantization mode of this layer, None if not quantized.
supports_masking	Whether this layer supports computing a mask using compute_mask.
trainable	Settable boolean, whether this layer should be trainable or not.
trainable_variables	List of all trainable layer state.
trainable_weights	List of all trainable weight variables of the layer.
variable_dtype	The dtype of the state (weights) of the layer.
variables	List of all layer state, including random seeds.
weights	List of all weight variables of the layer.

__init__(units, activation='elu', use_batch_norm=False, **kwargs)

Initialize the StaticEnrichmentLayer.

Parameters:

• units (int) – Number of hidden units in the internal GatedResidualNetwork.

• activation (str, optional) – Activation function for the GRN. Defaults to 'elu'.

• use_batch_norm (bool, optional) – Whether to apply batch normalization in the GRN. Defaults to False.

• **kwargs – Additional arguments passed to the parent Keras Layer.

call(temporal_features, context_vector, training=False)

Forward pass of the static enrichment layer.

Parameters:

• static_context_vector (tf.Tensor) – Static context of shape \((B, U)\).

• temporal_features (tf.Tensor) – Temporal features of shape \((B, T, D)\).

• training (bool, optional) – Whether the layer is in training mode. Defaults to False.

Returns:

Enriched temporal features of shape \((B, T, U)\), assuming units = U.

Return type:

tf.Tensor

Notes

1. Expand and tile static_context_vector over time steps.

2. Concatenate with temporal_features.

3. Pass through internal GRN for final transformation.

get_config()

Return the layer configuration for serialization.

Returns:

Configuration dictionary containing initialization parameters.

Return type:

dict

classmethod from_config(config)

Create a new instance from a config dictionary.

Parameters:

config (dict) – Configuration as returned by get_config.

Returns:

Instantiated layer object.

Return type:

StaticEnrichmentLayer

help(**kwargs)

my_params = StaticEnrichmentLayer(units, activation='elu', use_batch_norm=False)