arXiv：cs.LG· Junyi Yao, Zihao Zheng, Jiayu Long·· 4 小时前AI 评分32
最大似然成对排序的扰动敏感性研究：ASSA 攻击方法
Perturbation Sensitivity of Maximum-Likelihood Pairwise Ranking in Computational Decision Systems
AI 导读
研究提出自适应子集选择攻击（ASSA），将成对比较数据中的协同扰动建模为预算受限的子集选择问题，用于探测高影响扰动集。在合成与真实偏好数据集上，基于 MLE 的排序表现出明显的状态依赖性敏感：较小但协同的扰动即可引发输出排序的显著变化。ASSA 与随机、贪心及随机化子集基线在重复试验中进行了对比，刻画了扰动所致排序偏移的幅度与变异性。
正文
Abstract:Maximum-likelihood pairwise ranking is a com- mon computational mechanism for prioritization, reputation estimation, and comparison-driven decision support. Despite its broad use, the perturbation sensitivity of this estimator under structured changes in comparison data remains insufficiently characterized. We study this question as an applied-mathematics and computational-science problem in stability analysis. We for- mulate coordinated perturbation as a budgeted subset-selection problem over pairwise observations and introduce an Adaptive Subset Selection Attack (ASSA) as a scalable search heuristic for probing high-impact perturbation sets. Through experiments on synthetic and observed preference datasets, we show that MLE-based ranking can exhibit pronounced regime-dependent sensitivity: relatively small but coordinated perturbations may in- duce meaningful changes in output orderings, while the response profile varies across budgets and data conditions. By comparing ASSA with random, greedy, and randomized subset baselines under repeated trials, we characterize both the magnitude and the variability of perturbation-induced ranking shifts. These results position pairwise ranking sensitivity as a problem in computational reliability, numerical stability, and robustness auditing for engineering systems built on comparison-driven inference.
| Comments: | accepted to 2026 International Conference on Data Science, Mathematics, and Informatics (ICoDMI), proceedings to IEEE Xplore |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT) |
| Cite as: | arXiv:2604.17805 [cs.LG] |
| (or arXiv:2604.17805v5 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2604.17805 arXiv-issued DOI via DataCite |
Submission history
From: Junyi Yao [view email]
[v1]
Mon, 20 Apr 2026 04:52:30 UTC (258 KB)
[v2]
Mon, 15 Jun 2026 04:24:46 UTC (78 KB)
[v3]
Sun, 30 Aug 2026 19:21:03 UTC (79 KB)
[v4]
Tue, 15 Sep 2026 04:29:12 UTC (80 KB)
[v5]
Wed, 7 Oct 2026 04:23:50 UTC (81 KB)
来源：arXiv：cs.LG · arxiv.org