Semi-Supervised Predictive Clustering Trees for (Hierarchical) Multi-Label Classification

Table 6

Training times for SL-PCTs and SSL-PCTs.

Dataset	SL-PCT	SSL-PCT	CLUS-RF	SSL-RF
Hierarchical multi-label classification
Danish farms	0.3	3.8	1.4	1.4
Slovenian rivers	0.7	1.6	14.7	14.7
Enron	1	291.2	2.1	2.1
ImCLEF07A	0.4	20.1	2.4	2.4
ImCLEF07D	0.4	8.4	1.8	47.1
Diatoms	7.2	100.6	11.8	11.8
Cellcycle-GO	18.5	18.5	109.9	109.9
Church-GO	2.3	2.3	49.4	5
Derisi-GO	16.8	16.8	117.1	2500.4
Eisen-GO	12.1	196.8	157.3	157.3
Expr-GO	120.4	1031.6	267.3	267.3
Pheno-GO	1.2	1.2	4.8	4.8
Average	15.1	141.1	61.7	260.4
Multi-label classification
Bibtex	8.1	790.2	12.9	686.4
Birds	2.1	16.2	4.3	4.3
Corel5k	16.1	16.1	99.7	1314
Emotions	0.4	0.4	1.5	7.7
Enron	1.7	61.8	2.7	77.8
Genbase	0.1	0.5	0.4	0.4
Mediana	0.2	0.2	1.2	1.2
Medical	1	51.8	1.3	35.1
Scene	1.1	84.6	2.2	2.2
SIGMEA real	0.1	0.1	0.5	1
Slovenian rivers	0.2	0.4	1.4	3.8
Yeast	1	12	3.4	3.4
Average	2.7	86.2	11.0	178.1