基于梯度的解释在复杂性与忠实度之间的权衡
On the Complexity-Faithfulness Trade-off of Gradient-Based Explanations
AI总结：
针对ReLU网络梯度解释噪声大且现有平滑方法牺牲忠实度的问题，提出统一光谱框架量化平滑性与忠实度的权衡，揭示代理平滑导致的解释差距，并通过实验验证理论发现。
摘要：
ReLU网络在视觉数据中广泛应用，但其存在急剧过渡，有时依赖单个像素进行预测，导致原始基于梯度的解释充满噪声且难以解读。现有方法如GradCAM通过生成代理模型来平滑解释，却以牺牲忠实度为代价。我们提出一个统一的光谱框架，系统分析和量化解释的平滑性、忠实度及其权衡。利用该框架，我们量化并正则化ReLU网络对高频信息的贡献，提供识别这种权衡的原理性方法。我们的分析揭示了基于代理的平滑如何扭曲解释，导致“解释差距”，并针对不同事后方法正式定义和测量该差距。最后，我们在不同设计选择、数据集和消融实验中验证了理论发现。
英文摘要：
ReLU networks, while prevalent for visual data, have sharp transitions, sometimes relying on individual pixels for predictions, making vanilla gradient-based explanations noisy and difficult to interpret. Existing methods, such as GradCAM, smooth these explanations by producing surrogate models at the cost of faithfulness. We introduce a unifying spectral framework to systematically analyze and quantify smoothness, faithfulness, and their trade-off in explanations. Using this framework, we quantify and regularize the contribution of ReLU networks to high-frequency information, providing a principled approach to identifying this trade-off. Our analysis characterizes how surrogate-based smoothing distorts explanations, leading to an ``explanation gap'' that we formally define and measure for different post-hoc methods. Finally, we validate our theoretical findings across different design choices, datasets, and ablations.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
