fusionlab.nn.components.MultiResolutionAttentionFusion

class fusionlab.nn.components.MultiResolutionAttentionFusion

Bases: Layer, NNLearner

MultiResolutionAttentionFusion layer applying multi-head attention fusion over features [1].

This layer merges or fuses features at different resolutions or sources via multi-head attention. The input is projected to shape (B, T, D), and the output shares the same shape.

\[\mathbf{Z} = \text{MHA}(\mathbf{X}, \mathbf{X})\]

Parameters:

• units (int) – Dimension of the key, query, and value projections.

• num_heads (int) – Number of attention heads.

Notes

Typically used in multi-resolution contexts where time steps or multiple feature sets are merged.

call(`inputs`, training=False)

Forward pass of the multi-head attention layer.

get_config()

Returns config for serialization.

from_config(`config`)

Reconstructs the layer from a config.

Examples

>>> from fusionlab.nn.components import MultiResolutionAttentionFusion
>>> import tensorflow as tf
>>> x = tf.random.normal((32, 10, 64))
>>> # Instantiate multi-resolution attention
>>> mraf = MultiResolutionAttentionFusion(
...     units=64,
...     num_heads=4
... )
>>> # Forward pass => (32, 10, 64)
>>> y = mraf(x)

See also

HierarchicalAttention

Combines short and long-term sequences with attention.

ExplainableAttention

Another attention layer returning attention scores.

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)

Initialize the MultiResolutionAttentionFusion layer.

Parameters:

• units (int) – Dimensionality for the attention projections.

• num_heads (int) – Number of heads for multi-head attention.

Methods

__init__(units, num_heads)	Initialize the MultiResolutionAttentionFusion layer.
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(inputs[, training])	Forward pass applying multi-head attention to fuse features.
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)	Instantiate a new MultiResolutionAttentionFusion layer from config.
get_build_config()	Returns a dictionary with the layer's input shape.
get_config()	Returns configuration dictionary with 'units' and 'num_heads'.
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, num_heads)

Initialize the MultiResolutionAttentionFusion layer.

Parameters:

• units (int) – Dimensionality for the attention projections.

• num_heads (int) – Number of heads for multi-head attention.

call(inputs, training=False)

Forward pass applying multi-head attention to fuse features.

Parameters:

• inputs (tf.Tensor) – Tensor of shape (B, T, D).

• training (bool, optional) – Indicates training mode. Defaults to False.

Returns:

Tensor of shape (B, T, D), representing fused features.

Return type:

tf.Tensor

get_config()

Returns configuration dictionary with ‘units’ and ‘num_heads’.

Returns:

Configuration for serialization.

Return type:

dict

classmethod from_config(config)

Instantiate a new MultiResolutionAttentionFusion layer from config.

Parameters:

config (dict) – Configuration dictionary.

Returns:

A new instance of this layer.

Return type:

MultiResolutionAttentionFusion

help(**kwargs)

my_params = MultiResolutionAttentionFusion(units, num_heads)