samples with matching score below the updating thresh-old, as shown by the thinner edges (Fig. 5). However, the bottleneck represented by the edges related to samable 6 Statistics i.e., percentage of users affected by impostor intrusion and percentage of users not updated at all are computed for self-update. The operating threshold is set at 0.01% FAR and 1% FAR, for each biometrics (after multiple update cycles). For the purpose of comparison, percen-tage of users affected by impostor intrusion and percentage of users not updated at all are also shown for self-update and graph based min-cut technique.ple localized in sub-graph G3 (belonging to J2) is discarded for adaptation.