Dataset Musk 2

Basic characteristics Musk 2

6598

instances

102

bags

166

features

39

positive bags

between 1 and 1044

instances per bag

63

negative bags

Results Musk 2

AUC dataset 102
classifier error
AUC EER meanPrec avCost
APR 82.5 (1.2) 17.5 (1.2) 69.5 (2.0) 12.8 (0.7)
DiverseDensity (10.000000) 91.9 (1.3) 14.5 (1.9) 90.4 (1.4) 5.1 (0.7)
EMDD (10.000000) 86.9 (2.9) 19.9 (5.7) 84.4 (3.4) 6.9 (1.7)
MILBoost (100 rounds) 73.6 (4.5) 32.4 (4.1) 72.4 (3.9) 13.0 (1.5)
Citation 1NN, c=1 82.9 (1.2) 19.9 (1.6) 72.7 (0.9) 11.2 (0.4)
Citation 3NN, c=5 87.6 (1.7) 18.7 (1.4) 85.2 (1.7) 7.4 (0.6)
MI-SVM p=1.000000 81.5 (2.1) 29.2 (2.6) 80.9 (2.5) 10.2 (1.1)
MI-SVM r=10.000000 92.9 (1.4) 15.3 (2.4) 92.3 (1.7) 4.9 (1.1)
MILES p=1.000000 88.8 (1.8) 19.8 (4.9) 87.8 (1.4) 7.1 (0.7)
MILES r=10.000000 95.3 (1.5) 12.7 (2.8) 94.4 (1.9) 3.8 (0.8)
SVmil mahal.K 48.3 (5.7) 52.7 (9.1) 52.6 (3.9) 20.1 (0.8)
SVmil haussd.K 53.0 (32.4) 48.0 (26.3) 62.5 (20.7) 15.3 (7.2)
SVmil emd.K - - - -
SimpleMIL with Logistic2 80.4 (1.9) 28.9 (2.6) 79.2 (2.0) 10.5 (1.0)
mean-inst+Logistic2 89.5 (2.2) 18.4 (3.8) 85.8 (2.4) 6.9 (1.2)
extremes+Logistic2 89.3 (2.1) 21.1 (3.1) 86.5 (2.7) 7.3 (1.0)
cov-coef+Logistic2 82.1 (1.4) 25.0 (2.2) 75.4 (2.5) 11.0 (1.1)
Bag of Words (k=10)+Logistic2 65.2 (6.0) 38.4 (7.3) 64.4 (6.3) 15.7 (1.9)
Bag of Words (k=100)+Logistic2 74.9 (4.7) 32.7 (4.2) 70.3 (5.6) 12.9 (2.1)
PposteriorMIL 75.3 (4.4) 31.5 (4.7) 73.9 (5.5) 12.6 (2.2)
Dissim.repr. AUC dataset 102
classifier error
AUC EER meanPrec avCost
minmin.K+Logistic2 89.7 (2.1) 17.0 (2.2) 84.1 (3.6) 7.0 (1.2)
summin.K+Logistic2 56.9 (2.9) 46.5 (1.8) 59.4 (1.9) 17.3 (0.9)
meanmin.K+Logistic2 92.5 (1.8) 13.8 (1.9) 88.8 (3.4) 5.4 (1.2)
meanmean.K+Logistic2 83.3 (2.6) 20.9 (3.2) 74.3 (3.4) 10.8 (1.2)
mahal.K+Logistic2 52.2 (3.3) 47.5 (3.0) 43.4 (2.4) 23.5 (0.9)
haussd.K+Logistic2 92.7 (1.7) 11.8 (3.1) 87.4 (2.4) 5.2 (1.1)
emd.K+Logistic2 - - - -
linass.K+Logistic2 72.3 (7.6) 27.7 (7.8) 63.6 (6.5) 16.8 (2.6)