alpbench.benchmark.ActiveLearningSetting

Classes

ActiveLearningSetting(setting_id, ...)

Active Learning Setting

class alpbench.benchmark.ActiveLearningSetting.ActiveLearningSetting(setting_id, setting_name, setting_labeled_train_size, setting_train_type, setting_test_size, number_of_iterations, number_of_queries, factor)

Bases: object

Active Learning Setting

The active learning setting defines constraints and design choices of one active learning setup. This involves the size of the labeled training data, the size of the test data, the number of iterations, the number of samples queried per iteration, and a task-dependent factor in case a dynamic setting is considered (i.e. number of samples queried depends on the number of classes of the given dataset).

Parameters:

• setting_id (int) – id of the setting in the database

• setting_name (str) – descriptor of the setting

• setting_labeled_train_size (float) – size of the labeled training size

• setting_train_type (str) – type of the size parameter: number of data points or share of the (training) dataset

• setting_test_size (float) – size of the test data (always given as a share of the full dataset)

• number_of_iterations (int) – number of iterations

• number_of_queries (int) – number of queries queried per iteration

• factor (int) – task-dependent factor

setting_id

id of the setting in the database

Type:

int

setting_name

descriptor of the setting

Type:

str

setting_labeled_train_size

size of the labeled training size

Type:

float

setting_train_type

type of the size parameter: number of data points or share of the (training) dataset

Type:

str

setting_test_size

size of the test data (always given as a share of the full dataset)

Type:

float

number_of_iterations

number of iterations

Type:

int

number_of_queries

number of queries queried per iteration

Type:

int

factor

task-dependent factor

Type:

int

get_factor()

Get the factor.

get_number_of_iterations()

Get the number of iterations.

get_number_of_queries()

Get the number of samples queried per iteration.

get_setting_id()

Get the setting id.

get_setting_labeled_train_size()

Get the size of the labeled training data.

get_setting_name()

Get the setting name.

get_setting_test_size()

Get the size of the test data.

get_setting_train_type()

Get the training type, absolute or relative.