Are Classification Robustness and Explanation Robustness Really Strongly Correlated? An Analysis Through Input Loss Landscape

This paper looks into the critical area of deep learning robustness and challenges the common belief that classification robustness and explanation robustness in image classification systems are inherently correlated. Through a novel evaluation approach leveraging clustering for e!cient assessment of explanation robustness, we demonstrate that enhancing explanation robustness does not necessarily flatten the input loss landscape with respect to explanation loss - contrary to flattened loss landscapes indicating better classification robustness. To further investigate this contradiction, a training method designed to adjust the loss landscape with respect to explanation loss is proposed. Through the new training method, we uncover that although such adjustments can impact the robustness of explanations, they do not have an influence on the robustness of classification. These findings not only challenge the previous assumption of a strong correlation between the two forms of robustness but also pave new pathways for understanding the relationship between loss landscape and explanation loss. Codes are provided in the supplement.