torch_concepts.nn.functional.completeness_score

completeness_score(y_true, y_pred_blackbox, y_pred_whitebox, scorer=<function roc_auc_score>, average='macro')[source]

Calculate the completeness score for the given predictions and true labels. Main reference: “On Completeness-aware Concept-Based Explanations in Deep Neural Networks”

Parameters:

  • y_true (torch.Tensor) – True labels.

  • y_pred_blackbox (torch.Tensor) – Predictions from the blackbox model.

  • y_pred_whitebox (torch.Tensor) – Predictions from the whitebox model.

  • scorer (function) – Scoring function to evaluate predictions. Default is roc_auc_score.

  • average (str) – Type of averaging to use. Default is ‘macro’.

Returns:

Completeness score.

Return type:

float