Skip to main content

Table 1 Average number of true-positive and false-positive connections for MDL, best-fit-I, and best-fit-II filtered by CMI and MDL

From: Using the minimum description length principle to reduce the rate of false positives of best-fit algorithms

Noise (%) Algorithm m= 10 m= 20 m= 30 m= 40 m= 50
TP FP TP FP TP FP TP FP TP FP
0 MDL 10.9 3.0 15.4 1.1 17.0 0.5 17.5 0.3 17.7 0.1
BF-I 11.4 3.8 15.8 1.6 17.1 0.7 17.4 0.4 17.5 0.3
BF-I-CMI 10.4 3.2 14.8 1.3 15.9 0.6 16.2 0.4 16.3 0.3
BF-I-MDL 11.0 2.6 15.4 1.2 16.9 0.6 17.3 0.4 17.5 0.2
BF-II 11.7 2.8 16.1 1.5 17.3 0.7 17.6 0.6 17.7 0.3
BF-II-CMI 10.9 2.3 15.2 1.3 16.1 0.6 16.4 0.4 16.4 0.2
BF-II-MDL 10.8 1.9 15.3 0.9 16.9 0.4 17.5 0.3 17.6 0.2
5 MDL 9.5 5.8 14.1 5.8 16.2 5.5 17.0 5.9 17.4 6.4
BF-I 10.0 9.1 14.5 8.9 16.4 6.5 17.0 4.3 17.3 2.7
BF-I-CMI 9.1 6.7 13.5 7.1 15.2 5.2 15.7 3.8 15.9 2.5
BF-I-MDL 9.4 6.8 14.2 6.0 16.3 5.0 16.9 3.1 17.3 2.0
BF-II 10.4 7.3 14.9 8.5 16.6 6.8 17.3 4.6 17.5 3.0
BF-II-CMI 9.7 5.9 14.0 7.1 15.4 5.3 16.0 3.5 16.0 2.4
BF-II-MDL 9.3 4.9 14.0 5.3 16.2 4.7 17.0 3.4 17.3 2.2
10 MDL 8.3 8.1 12.8 10.4 15.1 10.6 16.2 10.7 16.9 11.0
BF-I 8.8 12.9 13.0 13.7 15.1 11.1 16.3 8.6 16.8 6.4
BF-I-CMI 7.9 9.4 12.1 11.0 13.9 9.7 14.9 7.7 15.3 4.5
BF-I-MDL 8.1 9.6 12.6 10.7 15.0 8.4 16.2 6.3 16.8 5.8
BF-II 9.2 10.9 13.5 13.1 15.6 11.4 16.6 9.2 17.1 7.0
BF-II-CMI 8.4 8.5 12.6 10.8 14.4 8.9 15.1 7.2 15.5 5.0
BF-II-MDL 8.1 7.5 12.6 9.0 15.1 8.5 16.3 6.9 16.9 5.6
  1. BF, best-fit.