平衡混合中的kurtosis对比的1/R定律
A 1/R Law for Kurtosis Contrast in Balanced Mixtures
AI总结:
本文提出平衡混合中kurtosis对比的1/R定律,并展示通过纯化方法恢复对比的机制,验证了理论预测和实验结果。
摘要:
基于kurtosis的独立成分分析(ICA)在宽泛且平衡的混合中效果减弱。我们证明了一个精确的冗余定律:对于一个标准化的投影,其有效宽度$R_{\mathrm{eff}}$(参与率),总体超额峰度服从$|κ(y)|=O(κ_{\max}/R_{\mathrm{eff}})$,在平衡情况下(通常$ c_b=O(\log R)$)得到紧致的$O(c_bκ_{\max}/R)$。作为不可能性筛选,标准有限矩条件下样本kurtosis估计,超越$O(1/\sqrt{T})$估计尺度需要$R\lesssim κ_{\max}\sqrt{T}$。我们还显示,纯化——选择$m\!\ll\!R$符号一致的源——恢复$R$独立的对比$Ω(1/m)$,并提供一个简单的数据驱动启发式方法。合成实验验证了预测的衰减、$\sqrt{T}$交叉点以及对比恢复。
英文摘要:
Kurtosis-based Independent Component Analysis (ICA) weakens in wide, balanced mixtures. We prove a sharp redundancy law: for a standardized projection with effective width $R_{\mathrm{eff}}$ (participation ratio), the population excess kurtosis obeys $|κ(y)|=O(κ_{\max}/R_{\mathrm{eff}})$, yielding the order-tight $O(c_bκ_{\max}/R)$ under balance (typically $c_b=O(\log R)$). As an impossibility screen, under standard finite-moment conditions for sample kurtosis estimation, surpassing the $O(1/\sqrt{T})$ estimation scale requires $R\lesssim κ_{\max}\sqrt{T}$. We also show that \emph{purification} -- selecting $m\!\ll\!R$ sign-consistent sources -- restores $R$-independent contrast $Ω(1/m)$, with a simple data-driven heuristic. Synthetic experiments validate the predicted decay, the $\sqrt{T}$ crossover, and contrast recovery.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。