Dataset Sival Banana

Basic characteristics Sival Banana

47414

instances

1500

bags

30

features

60

positive bags

between 31 and 32

instances per bag

1440

negative bags

Results Sival Banana

AUC dataset 132
classifier error
AUC EER meanPrec avCost
APR - - - -
DiverseDensity (10.000000) 59.5 (0.2) 42.0 (1.6) 11.5 (0.7) 19.6 (0.5)
EMDD (10.000000) 61.4 (0.0) 41.8 (0.0) 11.4 (0.0) 19.2 (0.0)
MILBoost (100 rounds) 89.5 (1.5) 19.1 (1.1) 48.3 (5.6) 8.7 (0.9)
Citation 1NN, c=1 62.3 (0.6) 38.0 (0.5) 15.9 (1.9) 21.2 (0.3)
Citation 3NN, c=5 70.4 (1.6) 32.4 (1.6) 29.4 (1.9) 18.2 (0.6)
MI-SVM p=1.000000 86.7 (1.9) 20.6 (3.5) 45.3 (4.9) 9.6 (1.2)
MI-SVM r=10.000000 83.1 (1.2) 23.7 (2.5) 38.1 (2.7) 11.6 (0.9)
MILES p=1.000000 - - - -
MILES r=10.000000 - - - -
SVmil mahal.K 43.3 (2.0) 55.8 (2.3) 6.8 (0.8) 22.8 (0.3)
SVmil haussd.K 70.1 (2.8) 33.6 (3.3) 16.5 (1.9) 16.5 (0.5)
SVmil emd.K 70.8 (3.0) 34.7 (2.5) 17.6 (1.9) 16.3 (0.6)
SimpleMIL with Logistic2 87.0 (0.3) 19.1 (1.2) 25.3 (1.9) 9.3 (0.3)
mean-inst+Logistic2 76.2 (1.1) 30.7 (1.5) 19.9 (1.3) 14.6 (0.7)
extremes+Logistic2 80.3 (1.4) 27.0 (1.9) 33.8 (1.4) 12.9 (0.4)
cov-coef+Logistic2 88.6 (0.6) 19.2 (0.9) 40.4 (2.5) 9.0 (0.5)
Bag of Words (k=10)+Logistic2 68.2 (3.1) 35.9 (2.7) 13.2 (2.4) 17.4 (1.2)
Bag of Words (k=100)+Logistic2 71.6 (2.6) 33.7 (2.6) 16.1 (2.9) 16.4 (1.0)
PposteriorMIL 63.0 (2.9) 39.5 (2.0) 11.8 (2.0) 18.6 (1.0)
mean-inst+LIBSVM 63.7 (3.6) 41.8 (3.6) 11.6 (1.9) 18.4 (0.7)
extremes+LIBSVM - - - -
cov-coef+LIBSVM - - - -
Bag of Words (k=10)+LIBSVM - - - -
Bag of Words (k=100)+LIBSVM - - - -

Results are obtained using 10-fold stratified crossvalidation. The best results are indicated in bold, together with the results that are not significantly worse (based on a paired-difference t-test with a confidence level of 0.05). The standard deviations are shown between brackets.

Dissim.repr. AUC dataset 132
classifier error
AUC EER meanPrec avCost
minmin.K+Logistic2 60.1 (0.3) 41.9 (1.6) 9.3 (0.6) 19.9 (0.1)
summin.K+Logistic2 - - - -
meanmin.K+Logistic2 - - - -
meanmean.K+Logistic2 - - - -
mahal.K+Logistic2 - - - -
haussd.K+Logistic2 - - - -
emd.K+Logistic2 - - - -
linass.K+Logistic2 - - - -

Results are obtained using 10-fold stratified crossvalidation. The best results are indicated in bold, together with the results that are not significantly worse (based on a paired-difference t-test with a confidence level of 0.05). The standard deviations are shown between brackets.