What happened
arXiv:2608.11285v1 Announce Type: new Abstract: Despite the practical relevance of sparse decision-based black-box threats, they have received limited attention in semantic segmentation. To bridge this gap, we adapt the
Analysis and impact
SegPAR：面向语义分割的类别中心稀疏决策攻击
研究问题与方法
语义分割任务需要对图像中每个像素进行分类，这使攻击面远大于图像分类。传统对抗攻击通常依赖梯度信息（白盒）或置信度分数（基于分数的黑盒），而决策型黑盒攻击只能观察模型的离散输出——即每个像素被分到哪个类别，无法获取任何概率或梯度信号。稀疏性约束进一步要求攻击者只能修改图像中极少量的像素，否则扰动容易被察觉或触发防御机制。
SegPAR 的核心思路是「类别中心」（class-centric）：原文摘要在此处截断，未提供方法的具体实现细节。从命名推断，该方法可能围绕目标类别在特征空间或决策边界上的代表性位置来设计和优化扰动，而非对每个像素独立搜索。但具体如何构造类别中心、如何选择被攻击的像素子集、以及优化过程如何运作，原文摘要未提供，需阅读全文才能确认。
证据强度与局限
作为一篇新发布的 arXiv 预印本（2026 年 8 月 13 日提交），该研究尚未经过同行评议。摘要仅提供了研究动机与方法名称，未披露实验数据集、对比基线、攻击成功率或扰动像素数量等关键量化指标。因此，目前无法评估其实际攻击效果与相对现有方法的优势幅度。读者应将此视为初步研究成果，其结论的可靠性有待正式发表后的验证。
意义与展望
如果 SegPAR 确实能在稀疏决策型约束下有效攻击语义分割模型，其实践意义值得关注：自动驾驶、医学影像分析、遥感监测等依赖语义分割的安全敏感场景，可能面临更隐蔽的对抗威胁——攻击者只需改变极少数像素且无需模型内部信息，就能诱导分割结果出错。这为相关系统的鲁棒性评估与防御设计提出了新的要求。不过，在缺乏完整实验数据的情况下，这一威胁的实际严重程度仍属未知。
References
Original source text
arXiv:2608.11285v1 Announce Type: new Abstract: Despite the practical relevance of sparse decision-based black-box threats, they have received limited attention in semantic segmentation. To bridge this gap, we adapt the most representative decision-based black-box sparse attacks from the classification domain to serve as baselines, establishing a rigorous benchmark for this underexplored setting. In this context, we demonstrate that one of the existing methods suffers from severe query inefficiency due to its image-centric pixel accumulation, which rapidly exhausts query budgets across the vast image space. To overcome this, we propose SegPAR, a novel decision-based framework that shifts to a class-centric exploration paradigm. Furthermore, to eliminate the misleading feedback generated by standard decision rewards during pixel accumulation, we introduce a novel discrepancy reward. Extensive experiments show that SegPAR significantly outperforms black-box baselines in sparsity efficiency and MIoU reduction, while remaining competitive with white-box sparse attacks. Code is available at {https://github.com/KAU-QuantumAILab/SegPAR}.