Dataset Biocreative component

Basic characteristics Biocreative component

13129

instances

718

bags

200

features

359

positive bags

between 1 and 53

instances per bag

359

negative bags

Results Biocreative component

AUC dataset 329
classifier error
AUC EER meanPrec avCost
APR - - - -
DiverseDensity (10.000000) 76.6 (0.2) 29.5 (0.6) 73.7 (0.6) 18.1 (0.1)
EMDD (10.000000) 76.1 (0.7) 30.0 (0.0) 73.3 (0.4) 18.3 (0.0)
MILBoost (100 rounds) 79.8 (0.9) 27.4 (0.8) 77.5 (1.5) 16.7 (0.3)
Citation 1NN, c=1 69.8 (0.0) 32.3 (0.0) 64.9 (0.0) 21.5 (0.0)
Citation 3NN, c=5 78.3 (0.0) 31.2 (0.0) 75.0 (0.0) 18.0 (0.0)
MI-SVM p=1.000000 83.8 (0.0) 26.3 (0.0) 81.9 (0.0) 15.1 (0.0)
MI-SVM r=10.000000 84.0 (0.0) 25.0 (0.0) 83.0 (0.0) 15.2 (0.0)
MILES p=1.000000 63.7 (0.0) 42.2 (0.0) 63.3 (0.0) 22.6 (0.0)
MILES r=10.000000 65.1 (0.0) 39.3 (0.0) 65.9 (0.0) 21.9 (0.0)
SVmil mahal.K 51.8 (10.1) 48.4 (8.5) 55.7 (4.1) 23.2 (1.2)
SVmil haussd.K 53.3 (18.0) 47.5 (13.7) 57.9 (6.9) 22.3 (2.3)
SVmil emd.K 55.9 (32.3) 45.3 (25.7) 65.1 (13.9) 20.1 (4.5)
SimpleMIL with Logistic2 81.9 (0.0) 25.3 (0.0) 78.2 (0.0) 15.5 (0.0)
mean-inst+Logistic2 80.1 (0.0) 29.7 (0.0) 76.2 (0.1) 16.2 (0.0)
extremes+Logistic2 76.8 (0.1) 29.7 (0.0) 72.5 (0.1) 17.8 (0.0)
cov-coef+Logistic2 58.7 (0.8) 42.2 (0.9) 56.2 (0.7) 24.1 (0.2)
Bag of Words (k=10)+Logistic2 73.7 (5.6) 32.5 (5.2) 70.0 (5.8) 18.8 (1.6)
Bag of Words (k=100)+Logistic2 78.4 (2.5) 28.4 (2.6) 75.8 (1.9) 17.7 (1.1)
PposteriorMIL 52.4 (3.5) 48.3 (2.3) 53.7 (5.3) 24.1 (1.3)
mean-inst+LIBSVM - - - -
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 329
classifier error
AUC EER meanPrec avCost
minmin.K+Logistic2 63.1 (0.8) 37.1 (0.7) 58.7 (0.7) 23.2 (0.2)
summin.K+Logistic2 62.3 (19.9) 41.6 (13.8) 65.6 (9.1) 21.2 (2.2)
meanmin.K+Logistic2 73.4 (0.6) 32.5 (1.3) 69.6 (0.5) 18.6 (0.3)
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.