Exploring adversarial attacks on deep neural networks (DNNs) is crucial for assessing and enhancing their adversarial robustness. Among various attack types, hard-label attacks that rely only on predicted labels offer a practical approach. This paper focuses on the challenging task of hard-label attacks within an extremely limited query budget, which is a significant achievement rarely accomplished by existing methods. To tackle this, we propose an attack framework that leverages geometric information from previous perturbation directions to form triangles and employs a heuristic perpendicular trial to effectively utilize the intermediate directions. Extensive experiments validate the effectiveness of our approach under strict query constraints and demonstrate its superiority to the state-of-the-art methods.

Deep neural networks often exhibit the overconfidence issue when encountering out-of-distribution (OOD) samples. To address this, leveraging large-scale pre-trained models like CLIP has shown promise. While CLIP has the capability to encode a vast array of interconnected concepts, current OOD detection methods based on it primarily focus on ID categories and a limited set of OOD categories. In this paper, we propose a novel approach that harnesses the power of WordNet to fully exploit the rich knowledge encapsulated within CLIP, resulting in enhanced OOD detection performance. Our methodology involves constructing a word tree that includes both in-distribution (ID) words and a large set of semantically similar OOD words selected from WordNet. By matching a test image with the concepts of the words in the word tree using CLIP, we estimate the probability of the image being classified as either ID or OOD. Furthermore, we introduce a conditional random field model to effectively handle both the parent-child and the sibling-sibling conflicts in the concept matching results. Extensive experiments under various ID/OOD settings demonstrate the effectiveness of our approach and its superiority over state-of-the-art methods.