Normalized AOPC: Fixing Misleading Faithfulness Metrics for Feature Attribution Explainability

Open in new window