跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.21973 · v1

How Many Persistent Features Does a Biometric Need? A Law Linking Equal Error Rate to Feature Count and Persistence

首次提交 2026-11-17 06:58
版本更新 2026-11-17 06:58
雷达观测 2026-11-18 19:43
arXiv ↗PDF ↗

摘要

A biometric designer controls the number of features and their temporal persistence. Earlier empirical work linked the equal error rate (EER) to persistence and to the required number of features by straight lines. We derive the law behind these regularities. For NN independent standardized Gaussian features with test-retest correlation ρρ, scored by cosine similarity, Fisher's zz-transform turns the genuine and impostor scores into two normals, giving EER=Φ(−12N atanh⁡ρ)\mathrm{EER}=Φ(-\frac{1}{2}\sqrt{N}\,\operatorname{atanh}ρ) and, inversely, N=(2q/atanh⁡ρ)2N=(2q/\operatorname{atanh}ρ)^2 with q=Φ−1(1−EER)q=Φ^{-1}(1-\mathrm{EER}); intercorrelated features enter through an effective count NeffN_{\mathrm{eff}}. The central extension lets persistence vary across comparisons with spread ττ on Fisher's scale: EER=Φ(−m/(τ2+v+v))\mathrm{EER}=Φ(-m/(\sqrt{τ^2+v}+\sqrt{v})) with v=1/(Neff−2)v=1/(N_{\mathrm{eff}}-2), which for a normal spread has a floor Φ(−m/τ)Φ(-m/τ) that no number of features can lower. With no fitted parameter the simple law reproduces published simulations and the EER-ICC line, and predicts the KCC-EER relation of eye-movement embeddings on GazeBase (R2=0.88R^2=0.88), though less well on GazeBaseVR (0.630.63). Read from the genuine and impostor scores, the general form matches the EER of eye movements, keystroke dynamics, the ECG and face embeddings over 101 conditions with R2=0.98R^2=0.98 and a median error of 0.60.6 points, against 0.870.87 and 1.91.9 points for the simple law. The 61 eye-movement, keystroke and ECG conditions check consistency on the same scores; the 40 face conditions use held-out subjects, with the same accuracy. The product τNeff−2τ\sqrt{N_{\mathrm{eff}}-2} gives a design rule: adding or decorrelating features pays fully while it is well below one and little once it exceeds about three, where only higher or more uniform persistence helps.

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

目前只收录了一个版本。收录下一版后,可以对照标题、作者、分类和摘要。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v1 · 2026-11-17 06:58 · 最新观测版本

How Many Persistent Features Does a Biometric Need? A Law Linking Equal Error Rate to Feature Count and Persistence

A biometric designer controls the number of features and their temporal persistence. Earlier empirical work linked the equal error rate (EER) to persistence and to the required number of features by straight lines. We derive the law behind these regularities. For NN independent standardized Gaussian features with test-retest correlation ρρ, scored by cosine similarity, Fisher's zz-transform turns the genuine and impostor scores into two normals, giving EER=Φ(−12N atanh⁡ρ)\mathrm{EER}=Φ(-\frac{1}{2}\sqrt{N}\,\operatorname{atanh}ρ) and, inversely, N=(2q/atanh⁡ρ)2N=(2q/\operatorname{atanh}ρ)^2 with q=Φ−1(1−EER)q=Φ^{-1}(1-\mathrm{EER}); intercorrelated features enter through an effective count NeffN_{\mathrm{eff}}. The central extension lets persistence vary across comparisons with spread ττ on Fisher's scale: EER=Φ(−m/(τ2+v+v))\mathrm{EER}=Φ(-m/(\sqrt{τ^2+v}+\sqrt{v})) with v=1/(Neff−2)v=1/(N_{\mathrm{eff}}-2), which for a normal spread has a floor Φ(−m/τ)Φ(-m/τ) that no number of features can lower. With no fitted parameter the simple law reproduces published simulations and the EER-ICC line, and predicts the KCC-EER relation of eye-movement embeddings on GazeBase (R2=0.88R^2=0.88), though less well on GazeBaseVR (0.630.63). Read from the genuine and impostor scores, the general form matches the EER of eye movements, keystroke dynamics, the ECG and face embeddings over 101 conditions with R2=0.98R^2=0.98 and a median error of 0.60.6 points, against 0.870.87 and 1.91.9 points for the simple law. The 61 eye-movement, keystroke and ECG conditions check consistency on the same scores; the 40 face conditions use held-out subjects, with the same accuracy. The product τNeff−2τ\sqrt{N_{\mathrm{eff}}-2} gives a design rule: adding or decorrelating features pays fully while it is well below one and little once it exceeds about three, where only higher or more uniform persistence helps.

查看此版本解读 →查看此版本原文 ↗