发表机构
H&K Research Studio, Clevix LLC; Banking Academy of Vietnam（H&K研究工作室，Clevix有限责任公司; 越南银行学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对独立训练神经网络因无共享参考框架及神经坍缩带来的坐标自由度问题，通过五个在MNIST上重建神经坍缩的网络，应用仿射校正对齐，验证了供体特异性功能指纹在特定测试下的可检测性。
AI 中文摘要
独立训练的神经网络没有共享的神经元索引参考框架，因此比较它们需要考虑坐标自由度。神经坍缩加剧了这个问题：网络趋向于共享的低维几何结构，这就引发了收敛后轨迹特定功能变化是否仍可区分的问题。我们区分了三个主张——可检测性、可移植性和因果持续性，并解决了第一个主张。使用五个在MNIST上独立训练以重建神经坍缩的网络，我们应用经过验证的仿射校正对齐将供体头部映射到受体坐标中。在受体级基线校正后，供体特异性功能指纹仍然可区分：所有20个有序的供体-受体对都被正确识别，精确排列p = 0.0083，对泄漏审核具有鲁棒性。这些发现在此处使用的测试下确立了可检测性，但未确立可移植性或因果持续性。该研究展示了对齐、模糊性诊断和泄漏控制如何在受控环境中结合起来测试跨网络变化；这是否能推广到其他情况尚待探讨。
英文摘要
Independently trained neural networks have no shared neuron-index reference frame, so comparing them requires accounting for coordinate freedom. Neural Collapse sharpens this problem: networks converge toward a shared, low-dimensional geometry, raising the question of whether trajectory-specific functional variation remains distinguishable after convergence. We distinguish three claims - detectability, transplantability, and causal persistence - and address the first. Using five independently trained networks reconstructing Neural Collapse on MNIST, we apply a verified affine-correct alignment mapping donor heads into recipient coordinates. Donor-specific functional fingerprints remain distinguishable after recipient-level baseline correction: all 20 ordered donor-recipient pairs are correctly identified, with an exact permutation p=0.0083, robust to a leakage audit. These findings establish detectability under the test used here, but not transplantability or causal persistence. The study shows how alignment, ambiguity diagnostics, and leakage control combine to test cross-network variation in a controlled setting; whether this generalizes beyond it is open.
Comments23 pages, 4 figures, 9 tables
