Open-set source tracing is increasingly framed as a verification problem, motivating the use of pairwise metric-learning objectives from biometrics. We thus compare global anchoring and pairwise verification under matched backbones and a fixed data and epoch budget on MLAAD (in-domain) and STOPA (out-of-domain). In our runs, global anchoring yields lower in-domain error (8.61% EER) than pairwise variants (12-15% EER), even with rival mining and XLS-R finetuning. Because pairwise objectives optimize similarity directly, they concentrate variance into fewer embedding directions, reducing resolution among closely related generators. To test if this drives the drop, we impose a similar bottleneck to the globally supervised baseline, yet the baseline remains competitive. Together with an embedding-space analysis ($k_{99}$), these results suggest that the gap is not explained by dimensionality alone, but rather by the pairwise objective's shaping of the retained directions.

翻译：开放集声源溯源日益被视为一个验证问题，这促使人们从生物特征识别中引入成对度量学习目标。因此，我们在匹配的主干网络以及固定的数据和轮次预算下，比较了全局锚定方法和成对验证方法在MLAAD（域内）和STOPA（域外）上的性能。在我们的实验中，即使采用对抗挖掘和XLS-R微调，全局锚定方法的域内错误率（8.61%等错误率）仍低于成对变体（12-15%等错误率）。由于成对目标直接优化相似性，它们将方差集中到更少的嵌入方向上，从而降低了密切相关生成器之间的分辨率。为了验证这是否是导致性能下降的原因，我们对全局监督基线施加了类似的瓶颈，但基线仍保持竞争力。结合嵌入空间分析（$k_{99}$），这些结果表明，差距并不能仅仅用维度来解释，而成对目标对保留方向的塑造才是关键因素。