Biomedical Text Categorization Based on Ensemble Pruning and Optimized Topic Modelling
Table 7
Comparison of the proposed text categorization scheme with conventional classifiers, ensemble learners, and ensemble pruning method (with BA-LDA (DB) based representation).
Classification algorithm
oh5
oh10
oh15
ohscal
ohsumed
NB
87.67
81.42
87.44
83.64
47.09
SVM
88.97
82.22
88.16
85.32
50.08
Bagging+NB
89.32
83.35
88.87
83.47
48.52
Bagging+SVM
88.03
84.84
87.86
83.92
50.73
AdaBoost+NB
89.77
83.60
87.48
86.18
51.18
AdaBoost+SVM
88.18
84.95
87.35
86.29
51.85
RandomSubspace+NB
88.32
83.96
86.66
88.09
50.70
RandomSubspace+SVM
88.56
84.11
89.58
88.29
50.29
Stacking
88.28
86.87
88.93
84.90
53.84
ESM
88.58
86.66
90.25
88.48
51.94
BES
89.29
86.00
90.98
89.12
52.47
LibD3C
90.35
87.95
91.27
90.48
53.41
CDM
91.51
89.61
93.17
91.33
54.47
Proposed scheme
93.14
91.29
93.76
92.14
58.17
NB: Naïve Bayes algorithm, SVM: support vector machines, ESM: ensemble selection from libraries of models, BES: Bagging ensemble selection, LibD3C: hybrid ensemble pruning based on k-means and dynamic selection, and CDM: ensemble pruning based on combined diversity measures.
