解释机器学习模型的预测是一项核心挑战，尤其是对于黑箱视觉模型而言。基于概念的归因方法采用人类可理解的概念来解释模型，现有的此类方法仅能解释分类和回归模型，且需要访问模型权重或带有大量概念标注的数据集，多数方法无法提供图像级归因。我们提出Hide and Seek（HnS），一种新的概念级归因方法，无需访问模型权重或带标注的概念数据集，可解释多种视觉模型。给定一个概念和图像集，HnS利用视觉语言模型检索与概念对齐的图像，在每张图像中定位该概念，对定位区域进行掩码处理，并通过黑箱模型在原始图像与掩码图像之间输出的变化来衡量归因，通过直接测量模型行为实现准确归因并支持局部归因。我们在CelebA和COCO-Stuff上对HnS进行了验证，与现有方法相比，HnS与真实归因的一致性更好。我们还展示了HnS在图像字幕、卫星图像分割等不同任务上的适用性，这些任务中没有现有概念归因方法可用。我们在补充材料中提供了匿名代码。
英文摘要：
Interpreting machine learning model predictions is a central challenge, particularly for black-box vision models. Concept-based attribution methods interpret models using human-understandable concepts. Existing concept-based methods can only interpret classification and regression models. They require access to model weights or large concept-annotated datasets, and most methods cannot provide image-level attribution. We propose Hide and Seek (HnS), a new method for concept-level attribution that does not require access to model weights or labeled concept datasets. HnS can interpret a wide range of vision models. Given a concept and image set, HnS retrieves concept-aligned images using a vision-language model, localizes the concept within each image, masks the localized region, and measures attribution from the change in the black-box model's output between the original and masked images. This enables faithful attribution by directly measuring the model's behavior and supporting local attributions. We validated HnS on CelebA and COCO-Stuff. Compared to existing methods, HnS showed better agreement with ground-truth attributions. We demonstrated the applicability of HnS on diverse tasks, including image captioning and satellite image segmentation, where no existing concept attribution method can be used. We provided our anonymous code in the supplementary material.
Hide and Seek in Embedding Space: Geometry-based Steganography and Detection in Large Language Models
UCL Centre for Artificial Intelligence, University College London, UK（伦敦大学学院人工智能中心，大学学院伦敦）；School of Computing and Communications, Lancaster University, UK（兰卡斯特大学计算机与通讯学院）；Department of Informatics, University of Sussex, UK（苏塞克斯大学信息学院）；Centre for Psychedelic Research and Centre for Complexity Science, Imperial College London, UK（伦敦帝国学院迷幻研究与复杂科学中心）；Centre for Eudaimonia and Human Flourishing, University of Oxford, UK（牛津大学幸福与人类繁荣中心）
We train embodied agents to play Visual Hide and Seek where a prey must navigate in a simulated environment in order to avoid capture from a predator. We place a variety of obstacles in the environment for the prey to hide behind, and we only give the agents partial observations of their environment using an egocentric perspective. Although we train the model to play this game from scratch, experiments and visualizations suggest that the agent learns to predict its own visibility in the environment. Furthermore, we quantitatively analyze how agent weaknesses, such as slower speed, effect the l…
We propose `Hide-and-Seek', a weakly-supervised framework that aims to improve object localization in images and action localization in videos. Most existing weakly-supervised methods localize only the most discriminative parts of an object rather than all relevant parts, which leads to suboptimal performance. Our key idea is to hide patches in a training image randomly, forcing the network to seek other relevant parts when the most discriminative part is hidden. Our approach only needs to modify the input image and can work with any network designed for object localization. During testing, we…
University of California, Davis（加州大学戴维斯分校）；Zhejiang University（浙江大学）
We propose 'Hide-and-Seek' a general purpose data augmentation technique, which is complementary to existing data augmentation techniques and is beneficial for various visual recognition tasks. The key idea is to hide patches in a training image randomly, in order to force the network to seek other relevant content when the most discriminative content is hidden. Our approach only needs to modify the input image and can work with any network to improve its performance. During testing, it does not need to hide any patches. The main advantage of Hide-and-Seek over existing data augmentation techn…
Image completion is the problem of generating whole images from fragments only. It encompasses inpainting (generating a patch given its surrounding), reverse inpainting/extrapolation (generating the periphery given the central patch) as well as colorization (generating one or several channels given other ones). In this paper, we employ a deep network to perform image completion, with adversarial training as well as perceptual and completion losses, and call it the ``missing data encoder'' (MDE). We consider several configurations based on how the seed fragments are chosen. We show that trainin…
Cortexica Vision Systems（科蒂西卡视觉系统公司）；Mullard Space Science Laboratory（穆拉德空间科学实验室）；University College London（伦敦大学学院）
In this paper, we examine the real-time recovery of a video tracker from a target loss, using information that is already available from the original tracker and without a significant computational overhead. More specifically, before using the tracker output to update the target position we estimate the detection confidence. In the case of a low confidence, the position update is rejected and the tracker passes to a single-frame failure mode, during which the patch low-level visual content is used to swiftly update the object position, before recovering from the target loss in the next frame. …
Hide-and-Seek Attribution: Weakly Supervised Segmentation of Vertebral Metastases in CT
Institute of Neuroradiology, TUM University Hospital, School of Medicine and Health, Technical University of Munich (TUM)（慕尼黑工业大学大学医院神经放射学研究所，医学与健康学院，慕尼黑工业大学）；Chair for AI in Healthcare and Medicine, TUM and TUM University Hospital（慕尼黑工业大学及大学医院医疗与医学人工智能教席）；Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心）；Institute of Diagnostic and Interventional Radiology, TUM University Hospital, School of Medicine and Health, Technical University of Munich (TUM)（慕尼黑工业大学大学医院诊断与介入放射学研究所，医学与健康学院，慕尼黑工业大学）；Department of Neuroradiology, University Hospital Munich (LMU)（慕尼黑大学医院神经放射学系）；Department of Quantitative Biomedicine, University of Zurich（苏黎世大学定量生物医学系）；Department of Computing, Imperial College London（伦敦帝国理工学院计算系）；Department of Diagnostic and Interventional Radiology, University Hospital Frankfurt（法兰克福大学医院诊断与介入放射学系）；Department of Orthopedics and Sports Orthopedics, TUM University Hospital, School of Medicine and Health, Technical University of Munich (TUM)（慕尼黑工业大学大学医院骨科与运动骨科，医学与健康学院，慕尼黑工业大学）
We present a novel approach for semantically targeted adversarial attacks on Optical Flow. In such attacks the goal is to corrupt the flow predictions of a specific object category or instance. Usually, an attacker seeks to hide the adversarial perturbations in the input. However, a quick scan of the output reveals the attack. In contrast, our method helps to hide the attackers intent in the output as well. We achieve this thanks to a regularization term that encourages off-target consistency. We perform extensive tests on leading optical flow models to demonstrate the benefits of our approach…
Council for Scientific and Industrial Research（科学与工业研究理事会）；University of Edinburgh（爱丁堡大学）；University of Pretoria（比勒陀利亚大学）
Saliency maps are often used in computer vision to provide intuitive interpretations of what input regions a model has used to produce a specific prediction. A number of approaches to saliency map generation are available, but most require access to model parameters. This work proposes an approach for saliency map generation for black-box models, where no access to model parameters is available, using a Bayesian optimisation sampling method. The approach aims to find the global salient image region responsible for a particular (black-box) model's prediction. This is achieved by a sampling-base…
Seoul National University（首尔大学）；ASRI（先进技术研究中心）；INMC（纳米材料与技术研究所）；ISRC（智能系统研究中心）；Institute of Engineering Research（工程研究院）
Weakly supervised segmentation methods using bounding box annotations focus on obtaining a pixel-level mask from each box containing an object. Existing methods typically depend on a class-agnostic mask generator, which operates on the low-level information intrinsic to an image. In this work, we utilize higher-level information from the behavior of a trained object detector, by seeking the smallest areas of the image from which the object detector produces almost the same result as it does from the whole image. These areas constitute a bounding-box attribution map (BBAM), which identifies the…
