f3dasm_optimize.Ftrl

class Ftrl(domain, learning_rate=0.001, learning_rate_power=-0.5, initial_accumulator_value=0.1, l1_regularization_strength=0.0, l2_regularization_strength=0.0, l2_shrinkage_regularization_strength=0.0, beta=0.0, **kwargs)

Bases: TensorflowOptimizer

List of Public Methods

update_step

Update step of the optimizer. Needs to be implemented

Attributes

require_gradients
type

Methods

_check_number_of_datapoints()

Check if the number of datapoints is sufficient

for the initial population

Raises:

ValueError – Raises when the number of datapoints is insufficient

_construct_model(data_generator)

Method that is called before the optimization starts. This method can be used to construct a model based on the available data or a specific data generator.

Parameters:

data_generator (DataGenerator) – DataGenerator object

Note

When this method is not implemented, the method will do nothing.

_get_info()

Give a list of characteristic features of this optimizer

Return type:

List[str]

Returns:

List of strings denoting the characteristics of this optimizer

property _population: int

Property to return the population size of the optimizer

Returns:

Number of individuals in the population

Return type:

int

Note

If the population is not set, the property will return 1 This is done to prevent errors when the population size is not an available attribute in a custom optimizer class.

_reset(data)

Reset the optimizer to its initial state

property _seed: int

Property to return the seed of the optimizer

Returns:

Seed of the optimizer

Return type:

int | None

Note

If the seed is not set, the property will return None This is done to prevent errors when the seed is not an available attribute in a custom optimizer class.

_set_algorithm()

Method that can be implemented to set the optimization algorithm. Whenever the reset method is called, this method will be called to reset the algorithm to its initial state.

_set_data(data)

Set the data attribute to the given data

update_step(data_generator)

Update step of the optimizer. Needs to be implemented

by the child class

Parameters:

data_generator (DataGenerator) – data generator object to calculate the objective value

Returns:

ExperimentData object containing the new samples

Return type:

ExperimentData

Raises:

NotImplementedError – Raises when the method is not implemented by the child class

Note

You can access the data attribute of the optimizer to get the available data points. The data attribute is an f3dasm.ExperimentData object.