RES: A Robust Framework for Guiding Visual Explanation
RES: A Robust Framework for Guiding Visual Explanation
复制标题
DOI：
10.1145/3534678.3539419
复制
发表时间：
2022-06
期刊：
影响因子：
--
通讯作者：
Yuyang Gao;Tong Sun;Guangji Bai;Siyi Gu;S. Hong;Liang Zhao
中科院分区：
文献类型：
--
作者：
Yuyang Gao;Tong Sun;Guangji Bai;Siyi Gu;S. Hong;Liang Zhao
尽管现代深度神经网络（DNN）中的解释技术进展迅速，主要焦点是处理“如何生成解释”，但检查解释本身质量的高级研究问题（例如，“解释是否准确”）并提高解释质量（例如，“当解释不准确时，如何调整模型以生成更准确的解释”）仍然相对未被探索。为了引导模型获得更好的解释，解释监督技术（在模型解释上添加监督信号）已经开始显示出对提高深度神经网络的泛化能力和内在可解释性的良好效果。然而，对监督解释的研究，特别是在通过显着图表示的基于视觉的应用中，由于几个固有的挑战而处于早期阶段：1）人类解释注释边界的不准确性，2）人类解释注释区域的不完整性，以及3）人类注释和模型解释图之间的数据分布的不一致性。为了应对这些挑战，我们提出了一个通用的RES框架，通过开发一个新的目标，处理不准确的边界，不完整的区域，和不一致的人类注释分布，指导视觉解释，与模型泛化的理论依据。在两个真实世界图像数据集上的大量实验证明了所提出的框架在增强骨干DNN模型的解释合理性和性能方面的有效性。
Despite the fast progress of explanation techniques in modern Deep Neural Networks (DNNs) where the main focus is handling "how to generate the explanations", advanced research questions that examine the quality of the explanation itself (e.g., "whether the explanations are accurate") and improve the explanation quality (e.g., "how to adjust the model to generate more accurate explanations when explanations are inaccurate") are still relatively under-explored. To guide the model toward better explanations, techniques in explanation supervision - which add supervision signals on the model explanation - have started to show promising effects on improving both the generalizability as and intrinsic interpretability of Deep Neural Networks. However, the research on supervising explanations, especially in vision-based applications represented through saliency maps, is in its early stage due to several inherent challenges: 1) inaccuracy of the human explanation annotation boundary, 2) incompleteness of the human explanation annotation region, and 3) inconsistency of the data distribution between human annotation and model explanation maps. To address the challenges, we propose a generic RES framework for guiding visual explanation by developing a novel objective that handles inaccurate boundary, incomplete region, and inconsistent distribution of human annotations, with a theoretical justification on model generalizability. Extensive experiments on two real-world image datasets demonstrate the effectiveness of the proposed framework on enhancing both the reasonability of the explanation and the performance of the backbone DNNs model.