Label Ranking Average Precision

Module Interface

class torchmetrics.classification.MultilabelRankingAveragePrecision(num_labels, ignore_index=None, validate_args=True, **kwargs)

Computes label ranking average precision score for multilabel data [1]. The score is the average over each ground truth label assigned to each sample of the ratio of true vs. total labels with lower score. Best score is 1.

As input to forward and update the metric accepts the following input:

• preds (Tensor): A float tensor of shape (N, C, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element.

• target (Tensor): An int tensor of shape (N, C, ...). Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if ignore_index is specified).

Note

Additional dimension ... will be flattened into the batch dimension.

As output to forward and compute the metric returns the following output:

• mlrap (Tensor): A tensor containing the multilabel ranking average precision.

Parameters

• num_labels (int) – Integer specifing the number of labels

• ignore_index (Optional[int]) – Specifies a target value that is ignored and does not contribute to the metric calculation

• validate_args (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.

Example

>>> from torchmetrics.classification import MultilabelRankingAveragePrecision
>>> _ = torch.manual_seed(42)
>>> preds = torch.rand(10, 5)
>>> target = torch.randint(2, (10, 5))
>>> mlrap = MultilabelRankingAveragePrecision(num_labels=5)
>>> mlrap(preds, target)
tensor(0.7744)

Initializes internal Module state, shared by both nn.Module and ScriptModule.

Functional Interface

torchmetrics.functional.classification.multilabel_ranking_average_precision(preds, target, num_labels, ignore_index=None, validate_args=True)

Computes label ranking average precision score for multilabel data [1]. The score is the average over each ground truth label assigned to each sample of the ratio of true vs. total labels with lower score. Best score is 1.

Accepts the following input tensors:

• preds (float tensor): (N, C, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element.

• target (int tensor): (N, C, ...). Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

Parameters

• preds (Tensor) – Tensor with predictions

• target (Tensor) – Tensor with true labels

• num_labels (int) – Integer specifing the number of labels

• ignore_index (Optional[int]) – Specifies a target value that is ignored and does not contribute to the metric calculation

• validate_args (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.

Example

>>> from torchmetrics.functional.classification import multilabel_ranking_average_precision
>>> _ = torch.manual_seed(42)
>>> preds = torch.rand(10, 5)
>>> target = torch.randint(2, (10, 5))
>>> multilabel_ranking_average_precision(preds, target, num_labels=5)
tensor(0.7744)

References

[1] Tsoumakas, G., Katakis, I., & Vlahavas, I. (2010). Mining multi-label data. In Data mining and knowledge discovery handbook (pp. 667-685). Springer US.

Return type

Tensor