fusionlab.nn.components.MultiDecoder

class fusionlab.nn.components.MultiDecoder

Bases: Layer, NNLearner

MultiDecoder for multi-horizon forecasting [1].

This layer takes a single feature vector per example of shape \((B, F)\) and produces a separate output for each horizon step, resulting in \((B, H, O)\).

\[\mathbf{Y}_h = \text{Dense}_h(\mathbf{x}),\, h \in [1..H]\]

Each horizon has its own decoder layer.

Parameters:

• output_dim (int) – Number of output features for each horizon.

• num_horizons (int) – Number of forecast horizons.

Notes

This layer is particularly useful when you want separate parameters for each horizon, instead of a single shared head.

call(`x`, training=False)

Forward pass that produces horizon-specific outputs.

get_config()

Returns configuration for serialization.

from_config(`config`)

Builds a new instance from config.

Examples

>>> from fusionlab.nn.components import MultiDecoder
>>> import tensorflow as tf
>>> # Input of shape (batch_size, feature_dim)
>>> x = tf.random.normal((32, 128))
>>> # Instantiate multi-horizon decoder
>>> decoder = MultiDecoder(output_dim=1, num_horizons=3)
>>> # Output shape => (32, 3, 1)
>>> y = decoder(x)

See also

MultiModalEmbedding

Provides feature embeddings that can be fed into MultiDecoder.

QuantileDistributionModeling

Projects deterministic outputs into multiple quantiles per horizon.

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__(output_dim, num_horizons)

Initialize the MultiDecoder.

Parameters:

• output_dim (int) – Number of features each horizon decoder should output.

• num_horizons (int) – Number of horizons to predict, each with its own Dense layer.

Methods

__init__(output_dim, num_horizons)	Initialize the MultiDecoder.
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(x[, training])	Forward pass: each horizon has a separate Dense 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 MultiDecoder from the config.
get_build_config()	Returns a dictionary with the layer's input shape.
get_config()	Returns 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__(output_dim, num_horizons)

Initialize the MultiDecoder.

Parameters:

• output_dim (int) – Number of features each horizon decoder should output.

• num_horizons (int) – Number of horizons to predict, each with its own Dense layer.

call(x, training=False)

Forward pass: each horizon has a separate Dense layer.

Parameters:

• x (tf.Tensor) – A 2D tensor (B, F).

• training (bool, optional) – Unused in this layer. Defaults to False.

Returns:

A 3D tensor of shape (B, H, O).

Return type:

tf.Tensor

get_config()

Returns layer configuration for serialization.

Returns:

Dictionary containing ‘output_dim’ and ‘num_horizons’.

Return type:

dict

classmethod from_config(config)

Create a new MultiDecoder from the config.

Parameters:

config (dict) – Contains ‘output_dim’, ‘num_horizons’.

Returns:

A new instance.

Return type:

MultiDecoder

help(**kwargs)

my_params = MultiDecoder(output_dim, num_horizons)