fusionlab.nn.components.HierarchicalAttention

class fusionlab.nn.components.HierarchicalAttention

Bases: Layer, NNLearner

Hierarchical Attention layer that processes short-term and long-term sequences separately using multi-head attention, then combines their outputs [1].

This allows the model to focus on different aspects of the data in short-term and long-term contexts and aggregate the attention outputs for a more comprehensive representation.

\[\mathbf{Z} = \text{MHA}(\mathbf{X}_{s}) + \text{MHA}(\mathbf{X}_{l})\]

where \(\mathbf{X}_{s}\) and \(\mathbf{X}_{l}\) are the short- and long-term sequences, respectively.

Parameters:

• units (int) – Dimensionality of the projection for the attention keys, queries, and values.

• num_heads (int) – Number of attention heads to use in each multi-head attention sub-layer.

Notes

The output shape depends on the last dimension in the short and long sequences, projected to units. The final output is the sum of the short-term attention output and the long-term attention output.

call(`inputs`, training=False)

Forward pass. Expects a list [short_term, long_term] with shapes (B, T, D_s) and (B, T, D_l).

get_config()

Returns configuration dictionary for serialization.

from_config(`config`)

Recreates the layer from a config dict.

Examples

>>> from fusionlab.nn.components import HierarchicalAttention
>>> import tensorflow as tf
>>> # Suppose short_term and long_term have
... # shape (batch_size, time_steps, features).
>>> short_term = tf.random.normal((32, 10, 64))
>>> long_term  = tf.random.normal((32, 10, 64))
>>> # Instantiate hierarchical attention
>>> ha = HierarchicalAttention(units=64, num_heads=4)
>>> # Forward pass
>>> outputs = ha([short_term, long_term])

See also

MultiModalEmbedding

Can precede attention by embedding multiple sources of input.

LearnedNormalization

Can be applied to short_term and long_term sequences prior to attention.

References

[1]

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017). “Attention is all you need.” In Advances in Neural Information Processing Systems (pp. 5998-6008).

__init__(units, num_heads)

Parameters:

• units (int)

• num_heads (int)

Methods

__init__(units, num_heads)
add_loss(losses, **kwargs)	Add loss tensor(s), potentially dependent on layer inputs.
add_metric(value[, name])	Adds metric tensor to the layer.
add_update(updates)	Add update op(s), potentially dependent on layer inputs.
add_variable(*args, **kwargs)	Deprecated, do NOT use! Alias for add_weight.
add_weight([name, shape, dtype, ...])	Adds a new variable to the layer.
build(input_shape)	Creates the variables of the layer (for subclass implementers).
build_from_config(config)	Builds the layer's states with the supplied config dict.
call(inputs[, training])	Forward pass of the HierarchicalAttention.
compute_mask(inputs[, mask])	Computes an output mask tensor.
compute_output_shape(input_shape)	Computes the output shape of the layer.
compute_output_signature(input_signature)	Compute the output tensor signature of the layer based on the inputs.
count_params()	Count the total number of scalars composing the weights.
finalize_state()	Finalizes the layers state after updating layer weights.
from_config(config)	Recreates the HierarchicalAttention layer from a config dictionary.
get_build_config()	Returns a dictionary with the layer's input shape.
get_config()	Returns a dictionary of config parameters for serialization.
get_input_at(node_index)	Retrieves the input tensor(s) of a layer at a given node.
get_input_mask_at(node_index)	Retrieves the input mask tensor(s) of a layer at a given node.
get_input_shape_at(node_index)	Retrieves the input shape(s) of a layer at a given node.
get_output_at(node_index)	Retrieves the output tensor(s) of a layer at a given node.
get_output_mask_at(node_index)	Retrieves the output mask tensor(s) of a layer at a given node.
get_output_shape_at(node_index)	Retrieves the output shape(s) of a layer at a given node.
get_params([deep])	Get the parameters for this learner.
get_weights()	Returns the current weights of the layer, as 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.
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 weights of the layer, from NumPy arrays.
summary()	Provide a summary of the learner's parameters.
with_name_scope(method)	Decorator to automatically enter the module name scope.

Attributes

activity_regularizer	Optional regularizer function for the output of this layer.
compute_dtype	The dtype of the layer's computations.
dtype	The dtype of the layer weights.
dtype_policy	The dtype policy associated with this layer.
dynamic	Whether the layer is dynamic (eager-only); set in the constructor.
inbound_nodes	Return Functional API nodes upstream of this layer.
input	Retrieves the input tensor(s) of a layer.
input_mask	Retrieves the input mask tensor(s) of a layer.
input_shape	Retrieves the input shape(s) of a layer.
input_spec	InputSpec instance(s) describing the input format for this layer.
losses	List of losses added using the add_loss() API.
metrics	List of metrics attached to the layer.
my_params
name	Name of the layer (string), set in the constructor.
name_scope	Returns a tf.name_scope instance for this class.
non_trainable_variables	Sequence of non-trainable variables owned by this module and its submodules.
non_trainable_weights	List of all non-trainable weights tracked by this layer.
outbound_nodes	Return Functional API nodes downstream of this layer.
output	Retrieves the output tensor(s) of a layer.
output_mask	Retrieves the output mask tensor(s) of a layer.
output_shape	Retrieves the output shape(s) of a layer.
stateful
submodules	Sequence of all sub-modules.
supports_masking	Whether this layer supports computing a mask using compute_mask.
trainable
trainable_variables	Sequence of trainable variables owned by this module and its submodules.
trainable_weights	List of all trainable weights tracked by this layer.
updates
variable_dtype	Alias of Layer.dtype, the dtype of the weights.
variables	Returns the list of all layer variables/weights.
weights	Returns the list of all layer variables/weights.

__init__(units, num_heads)

Parameters:

• units (int)

• num_heads (int)

call(inputs, training=False)

Forward pass of the HierarchicalAttention.

Parameters:

• inputs (list of tf.Tensor) – A list [short_term, long_term]. Each tensor should have shape \((B, T, D)\).

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

Returns:

A tensor of shape \((B, T, U)\), where U = units, representing the combined attention outputs.

Return type:

tf.Tensor

get_config()

Returns a dictionary of config parameters for serialization.

Returns:

Dictionary with ‘units’, ‘short_term_dense’ config, and ‘long_term_dense’ config.

Return type:

dict

classmethod from_config(config)

Recreates the HierarchicalAttention layer from a config dictionary.

Parameters:

config (dict) – Configuration dictionary.

Returns:

A new instance with the specified configuration.

Return type:

HierarchicalAttention

help(**kwargs)

my_params = HierarchicalAttention(units, num_heads)