Skip to main content
. 2018 Jun 27;34(13):i323–i332. doi: 10.1093/bioinformatics/bty252

Table 3.

Micro-average performance and computation time (for prediction) of kernel-based methods in Shen et al. (2014) and proposed methods in this paper

Method Acc F1 score Run. time
(%) (%) (ms)
PPK (Peaks) 75.74 (±6.72) 60.59 (±14.54) 52.37
LB (Loss binary) 76.63 (±7.03) 61.64 (±15.48) 1501.02
LC (Loss count) 75.33 (±5.4) 61.25 (±13.99) 1501.02
LI (Loss intensity) 74.54 (±8.49) 58.46 (±16.01) 1501.02
NB (Node binary) 79.11 (±5.02) 67.34 (±11.75) 1501.09
NI (Node intensity) 78.41 (±4.99) 66.87 (±12.11) 1501.01
CPC (Common path count) 79.02 (±7.4) 67.55 (±12.93) 1501.11
ComFT (combining all above) 80.98 (±6.05) 69.04 (±11.98) 1559.20
ComALIGNF (Proposed: MKL) 79.03 (±7.89) 65.67 (±13.02) 471.71
SIMPLE (Proposed) 78.33 (±6.05) 66.70 (±13.03) 4.57
L-SIMPLE (Proposed) 78.86 (±5.87) 67.59 (±12.35) 4.32