Discriminative Extreme Learning Machine with Cross-Domain Mean Approximation for Unsupervised Domain Adaptation

Table 3

Classification accuracies (%) of all algorithms on office + caltech and USPS + MNIST.

Methods	Data set
	C ⟶ A (1)	C ⟶ W (2)	C ⟶ D (3)	A ⟶ C (4)	A ⟶ W (5)	A ⟶ D (6)	W ⟶ C (7)	W ⟶ A (8)	W ⟶ D (9)	D ⟶ C (10)	D ⟶ A (11)	D ⟶ W (12)	USPS ⟶ MNIST	MNIST ⟶ USPS	Average
1NN	23.7	25.8	25.5	26.0	29.8	25.5	19.9	23.0	59.2	26.3	28.5	63.4	44.7	65.9	34.8
SVM	52.4	40.7	42.0	44.1	41.7	40.8	29.9	34.1	87.9	31.4	32.4	73.6	35.1	54.7	45.8
ELM	50.6	42.7	42.0	43.0	37.6	42.7	32.2	36.9	80.3	30.1	31.5	71.9	30.7	51.4	44.5
SSELM	52.9	45.4	38.9	41.8	35.9	38.9	34.3	37.1	79.0	30.4	32.1	80.0	29.1	43.3	44.2
TCA1	55.9	48.8	51.6	44.2	40.3	36.9	32.3	38.4	81.5	36.2	39.4	82.4	35.6	53.3	48.3
TCA2	55.0	48.5	49.7	44.4	41.0	43.3	32.6	38.9	79.6	34.6	38.3	83.9	34.7	52.8	48.4
JDA1	53.3	46.8	52.9	42.7	36.3	40.8	33.7	47.6	86.9	36.5	43.3	82.7	34.8	54.9	49.5
JDA2	54.8	51.9	51.2	41.2	39.7	44.6	35.1	45.5	80.3	33.4	40.9	83.1	33.9	53.1	49.2
DAELM_S	50.3	45.4	37.6	42.9	38.3	36.9	34.2	36.4	80.9	32.3	34.4	72.9	44.4	47.1	45.3
DAELM_T	27.9	31.9	22.9	25.3	19.3	36.3	41.9	46.5	53.5	48.1	51.2	54.2	31.1	60.3	39.3
KMM	48.4	45.8	53.6	42.3	42.5	42.7	31.6	32.0	72.0	31.7	32.2	42.9	33.0	42.0	42.3
LMPROJ	52.5	46.9	40.8	43.6	38.6	38.2	30.6	36.4	83.5	30.7	34.0	81.7	46.3	57.7	47.2
ARRLS	54.9	50.5	44.6	42.5	39.0	38.9	34.7	39.9	78.3	32.2	37.2	82.7	34.6	52.5	47.3
TELM-OWA	31.7	32.2	33.1	38.9	38.6	43.3	33.5	38.0	77.1	32.5	36.2	78.6	59.6	59.4	45.2
CDELM-M	52.1	51.1	45.9	42.3	42.9	45.9	30.3	39.8	81.2	30.3	35.7	81.8	46.5	62.5	49.2
CDELM-C	56.3	51.0	43.8	43.2	40.8	41.8	34.4	39.6	83.1	34.4	39.5	84.3	45.9	42.6	48.6
JUC-SDELM	49.6	39.8	44.5	42.5	39.0	44.1	29.7	35.6	81.5	27.7	36.4	74.6	39.8	62.7	46.2
ELM-CDMA	53.2	47.1	45.2	43.6	44.1	47.1	33.1	36.6	84.1	33.1	36.5	74.6	45.3	82.6	50.5
DELM-CDMA	55.3	48.5	45.9	42.2	44.8	45.2	33.9	37.3	84.1	34.0	38.1	81.0	47.5	84.4	51.6
DKELM-CDMA	57.9	56.6	59.9	47.4	45.8	46.5	38.7	41.4	93.6	38.0	42.8	89.5	55.6	79.8	56.7