torcheval.metrics.functional.multilabel_recall_at_fixed_precision

torcheval.metrics.functional.multilabel_recall_at_fixed_precision(input: Tensor, target: Tensor, *, num_labels: int, min_precision: float) → Tuple[List[Tensor], List[Tensor]]

Returns the highest possible recall value give the minimum precision for each label and their corresponding thresholds for multi-label classification tasks. The maximum recall computation for each label is equivalent to _binary_recall_at_fixed_precision_compute in binary_recall_at_fixed_precision.

Its class version is torcheval.metrics.MultilabelRecallAtFixedPrecision. See also binary_recall_at_fixed_precision

Parameters:

• input (Tensor) – Tensor of label predictions It should be probabilities with shape of (n_samples, n_label)
• target (Tensor) – Tensor of ground truth labels with shape of (n_samples, n_label)
• num_labels (int) – Number of labels
• min_precision (float) – Minimum precision threshold

Returns:

List[torch.Tensor], thresholds: List[torch.Tensor])

recall: List of max recall values for each label thresholds: List of best threshold values for each label

Return type:

a tuple of (recall

Examples

>>> import torch
>>> from torcheval.metrics.functional import multilabel_recall_at_fixed_precision
>>> input = torch.tensor([[0.75, 0.05, 0.35], [0.45, 0.75, 0.05], [0.05, 0.55, 0.75], [0.05, 0.65, 0.05]])
>>> target = torch.tensor([[1, 0, 1], [0, 0, 0], [0, 1, 1], [1, 1, 1]])
>>> multilabel_recall_at_fixed_precision(input, target, num_labels=3, min_precision=0.5)
([tensor([1.0, 1.0, 1.0], tensor([0.05, 0.55, 0.05])])