BriefGPT.xyz
Dec, 2023
用对抗方法估计拓扑维度的谐波评分图
Adversarial Estimation of Topological Dimension with Harmonic Score Maps
HTML
PDF
Eric Yeats, Cameron Darwin, Frank Liu, Hai Li
TL;DR
利用对得分映射的固定点 insight，通过利用对应的狄利克雷能量进行正则化将其前提条件确认为对潜在 manifold 学习的拓扑维度进行回归是可能的，然后引入了一种使用对抗性攻击来测量学习的 manifold 的拓扑维度的新方法，从而产生对学习的 manifold 的有用解释。
Abstract
quantification
of the number of variables needed to locally explain
complex data
is often the first step to better understanding it. Existing techniques from
→