SHIELD: A Regularization Technique to Enhance Explainability and Performance of Artificial Intelligence Models

SHIELD regularization aims to improve model explainability and performance by selectively concealing input features and assessing the resulting discrepancy in predictions.