Skip to main content

Table 11 EER, FRR 1% ,FRR 10% , AUC, HTER and Rank-1/5 identification rates of quality-independent and quality-dependent approaches for Protocol 2

From: MultiQ: single sensor-based multi-quality multi-modal large-scale biometric score database and its performance evaluation

Fusion rule Training set EER [%] FRR 1% [%] FRR 10% [%] AUC [%] HTER [%] Rank-1 [%] Rank-5 [%]
Sum Q H H85 33.2 74.5 50.4 23.3 41.2 38.5 50.6
GMM   26.4 76.4 40.1 19.5 32.5 26.0 38.7
LLR   25.8 71.1 43.5 16.8 34.2 41.8 56.1
RankSVM   28.8 72.2 46.1 18.4 37.2 40.9 54.4
GMM (Q-stack) Set 1 49.9 99.0 89.9 49.9 49.9 0.1 0.5
LLR (Q-stack)   13.0 69.8 18.9 5.5 12.5 31.1 48.1
RankSVM (Q-stack)   20.9 68.4 37.4 11.0 26.1 38.7 54.6
  1. FRR 1% and FRR 10% are FRR at 1% FAR and FRR at 10% FAR respectively