Skip to main content
. 2015 Nov 4;16:366. doi: 10.1186/s12859-015-0768-9

Fig. 5.

Fig. 5

The attenuation factor, delta(δ) and ISNCA convergence properties. The selection of attenuation factor, δ with iterations is presented for five different scenarios (a) and four other scenarios (b). The corresponding two lower panels shows the error convergence responses to the ISNCA algorithm with ROBNCA, for the different δ values. a The δ is fixed at 0.5 at iteration one (k = 1) in all scenarios and δ is varied (0.2, 0.5, 0.8, 1.0 and 1.2) at subsequent iterations, k >1. The δ (k = 1) = 0.5 and δ (k >1) = 1.0 gives the best error and smooth convergence. b) The δ is fixed at 1.0 at all iterations (k >1) in all scenarios and δ is varied (0.3, 0.5, 0.7 and 1.0) at first iteration. The delta values δ (k = 1) = 0.5 and δ (k >1) = 1.0 gives the best error and smooth convergence (red solid line)