发表机构
Human Technology Institute; University of Technology Sydney（人文技术研究所; 悉尼科技大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
Hide&Seek是一种端到端可微的逐实例特征选择模型，通过将特征去除转化为可微操作并结合简约权重退火框架，解决了现有方法的信息泄露和训练缓慢问题，性能优于现有最先进模型。
AI 中文摘要
逐实例特征选择是用于解释带标签数据及黑盒模型预测的重要工具。与全局特征选择技术不同，逐实例方法会为每个实例动态识别重要特征。越来越多的方法会学习一个选择器（用于识别重要特征）和一个预测器（利用这些特征进行预测）。然而，这些开创性方法面临信息泄露和缺乏可微性的挑战，这会减缓训练速度。本文提出了Hide&Seek，一种用于逐实例特征选择的端到端可微模型。我们在单一目标下联合学习特征选择与预测，且无信息泄露。Hide&Seek在一系列实验中优于现有最先进模型，且训练速度快。我们通过将特征去除重新表述为可微操作来实现这一点：不是离散地去除特征，而是替换每个特征的一定比例。此外，我们通过简约权重退火框架进一步稳定训练。
英文摘要
Instance-wise feature selection is a valuable tool for interpreting labeled data and the predictions of black-box models. In contrast to global feature selection techniques, instance-wise methods dynamically identify important features for each instance. A growing number of methods learn a selector, which identifies important features, and a predictor, which uses these to make predictions. However, these pioneering methods face challenges including information leakage and lack of differentiability, which can slow training. In this paper, we present Hide&Seek, an end-to-end differentiable model for instance-wise feature selection. We jointly learn feature selection and prediction under a single objective without information leakage. Hide&Seek outperforms existing state-of-the-art models across a range of experiments and is fast to train. We achieve this by reformulating feature removal as a differentiable operation where instead of discretely removing features, we replace a proportion of each feature. Training is further stabilized via a parsimony-weight annealing framework.
Comments27 pages, 12 figures, 16 tables. Accepted at ICML 2026 (PMLR 306). Code at https://github.com/talellinson/hide-and-seek-icml2026
