Due to the black-box nature of deep networks. making explanations of their decision-making is extremely challenging. A solution is using post-hoc attention mechanisms with the deep network to verify the decision basis. However. https://www.johnperformance.com/product-category/pretend-play/