从像素到感知：通过逐实例分组特征选择实现可解释预测
From Pixels to Perception: Interpretable Predictions via Instance-wise Grouped Feature Selection
AI总结：
针对机器学习模型决策可解释性问题，提出一种基于语义区域逐实例稀疏化、可动态调整稀疏度的可解释预测方法，在半合成与自然图像数据集上表现优于现有最优基准。
摘要：
理解机器学习模型的决策过程，能为任务、数据以及模型失效的背后原因提供宝贵洞见。本文提出一种方法，通过对输入图像进行逐实例稀疏化，实现本质上可解释的预测。为让稀疏化过程契合人类感知，我们在具有语义意义的像素区域空间而非像素层面学习掩码。此外，我们提出一种显式方法，可动态确定每个实例所需的稀疏程度。我们在半合成与自然图像数据集上的实证表明，该本质可解释分类器比现有最优基准能产生更有意义、更易被人类理解的预测。
英文摘要：
Understanding the decision-making process of machine learning models provides valuable insights into the task, the data, and the reasons behind a model's failures. In this work, we propose a method that performs inherently interpretable predictions through the instance-wise sparsification of input images. To align the sparsification with human perception, we learn the masking in the space of semantically meaningful pixel regions rather than on pixel-level. Additionally, we introduce an explicit way to dynamically determine the required level of sparsity for each instance. We show empirically on semi-synthetic and natural image datasets that our inherently interpretable classifier produces more meaningful, human-understandable predictions than state-of-the-art benchmarks.
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。
