incerto.active.DiversitySampling

incerto.active.DiversitySampling#

class incerto.active.DiversitySampling(acquisition_fn, batch_size=100, diversity_weight=0.5)[source]#

Bases: object

Diversity-based sampling with uncertainty.

Balances uncertainty with diversity to avoid selecting redundant samples.

Reference:

Brinker, “Incorporating Diversity in Active Learning” (ICML 2003)

Parameters:

  • acquisition_fn (BaseAcquisition) – Acquisition function

  • batch_size (int) – Number of samples to select

  • diversity_weight (float) – Weight for diversity term (0-1)

__init__(acquisition_fn, batch_size=100, diversity_weight=0.5)[source]#
Parameters:

  • acquisition_fn (BaseAcquisition)

  • batch_size (int)

  • diversity_weight (float)

Methods

__init__(acquisition_fn[, batch_size, ...])

query(model, x_unlabeled[, features])

Query samples balancing uncertainty and diversity.
__init__(acquisition_fn, batch_size=100, diversity_weight=0.5)[source]#
Parameters:

  • acquisition_fn (BaseAcquisition)

  • batch_size (int)

  • diversity_weight (float)

query(model, x_unlabeled, features=None)[source]#

Query samples balancing uncertainty and diversity.

Parameters:

  • model (Module) – Trained model

  • x_unlabeled (Tensor) – Unlabeled data

  • features (Optional[Tensor]) – Optional precomputed features for diversity computation

Return type:

Tensor

Returns:

Indices of selected samples

Contents