Combining multiple hypothesis testing and affinity propagation clustering leads to accurate, robust and sample size independent classification on gene expression data

Table 6 An overview of the published classification results in Golub et al. ALL/AML leukemia data

Authors	Cross Validation		Train-Test		Features
	Samples	Accuracy (%)	Samples	Accuracy (%)
[6]	36/38	94.73	29/34	85.29	50
[67]	38/38	100.00	34/34	100	-
[67]	-	-	33/34	97.06	-
[64]	-	-	-	94.1	-
[66]	-	-	-	94.1	-
[68]	-	-	-	91.6	-
[68]	-	-	-	94.4	-
[68]	-	-	-	95.8	-
[65]	-	-	-	94.17	50
[65]	-	-	-	95.44	50
[65]	-	-	-	95.94	50
[65]	-	-	-	96.44	50
[70]	38/38	100	31/34	91.17	7129
[70]	38/38	100	34/34	100	999
[70]	38/38	100	32/34	94.11	99
[70]	38/38	100	30/34	88.23	49
[70]	-	-	34/34	100	40
[70]	-	-	32/34	94.11	5
[71]	-	-	-	95.0~	-
[71]	-	-	-	95.0~	-
[71]	-	-	-	95.0~	-
[72]	37/38	98	34/34	100	185
[73]	38/38	100	34/34	100	3800
[74]	37/38	98	32/34	94.11	21
[62]	71/72	98.6	-	-	-
[38]	71/72	98.61	-	-	2
[69]	38/38 (DLDA)	100	33/34 (DLDA)	97.06	50
[60]	38/38	100	-	-	50
[63]	-	-	31/34	91.18	1038
mAP-KL (RF)	-	98.93	24/34	70.59	5
mAP-KL (KNN)	-	93.61	24/34	70.59	5
mAP-KL (SVM-linear)	-	97.36	27/34	79.41	5

ISSN: 1471-2105