Research Article

Robust Proximal Support Vector Regression Based on Maximum Correntropy Criterion

Table 1

Experiment results on .

NoiseAlgorithm()RMSEMAESSE/SSTSSR/SST

Type ALSSVR()0.0863(5.5)0.0676(5)0.0720(5)1.0761(2)
PSVR()0.0863(5.5)0.0677(6)0.0721(6)1.0762(1)
WLSSVR-H()0.0422(3)0.0349(3)0.0181(3)1.0344(6)
WLSSVR-L()0.0514(4)0.0408(4)0.0262(4)1.0479(3)
RLSSVR-MCC()0.0409(2)0.0331(1.5)0.0173(1.5)1.0446(4.5)
RPSVR-MCC()0.0408(1)0.0331(1.5)0.0173(1.5)1.0446(4.5)

Type BLSSVR()0.0979(5)0.0795(5)0.0944(5)0.9374(6)
PSVR()0.0980(6)0.0796(6)0.0945(6)0.9378(5)
WLSSVR-H()0.0723(3)0.0599(3)0.0503(3)0.9976(4)
WLSSVR-L()0.0771(4)0.0642(4)0.0588(4)1.0497(1)
RLSSVR-MCC()0.0694(1.5)0.0580(1.5)0.0464(1.5)1.0399(2)
RPSVR-MCC()0.0694(1.5)0.0580(1.5)0.0464(1.5)1.0398(3)

Type CLSSVR()0.0828(5.5)0.0665(6)0.0664(5.5)1.1137(1)
PSVR()0.0828(5.5)0.0664(5)0.0664(5.5)1.1136(2)
WLSSVR-H()0.0289(3)0.0221(3)0.0083(3)0.9816(4)
WLSSVR-L()0.0378(4)0.0308(4)0.0142(4)1.0154(3)
RLSSVR-MCC()0.0276(2)0.0213(1.5)0.0075(1.5)0.9737(5)
RPSVR-MCC()0.0275(1)0.0213(1.5)0.0075(1.5)0.9736(6)

Type DLSSVR()0.0899(5)0.0707(5)0.0772(5)0.9488(6)
PSVR()0.0900(6)0.0708(6)0.0774(6)0.9492(5)
WLSSVR-H()0.0531(1)0.0424(1)0.0287(1)1.0116(2)
WLSSVR-L()0.0610(4)0.0500(4)0.0357(4)1.0495(1)
RLSSVR-MCC()0.0546(2.5)0.0443(2.5)0.0293(2.5)0.9703(4)
RPSVR-MCC()0.0546(2.5)0.0443(2.5)0.0293(2.5)0.9704(3)

Type ELSSVR()0.2120(6)0.1749(6)0.4268(6)0.7142(4)
PSVR()0.2109(5)0.1742(5)0.4226(5)0.7110(5)
WLSSVR-H()0.2004(4)0.1636(4)0.3836(4)0.5573(6)
WLSSVR-L()0.1887(3)0.1552(2)0.3434(3)0.7500(3)
RLSSVR-MCC()0.1848(2)0.1553(3)0.3412(2)0.7873(2)
RPSVR-MCC()0.1845(1)0.1548(1)0.3405(1)0.7881(1)

Type FLSSVR()0.1796(5)0.1498(5)0.3112(5)0.9709(2)
PSVR()0.1791(4)0.1493(4)0.3091(4)0.9699(3)
WLSSVR-H()0.1846(6)0.1507(6)0.3270(6)0.8896(6)
WLSSVR-L()0.1730(1)0.1406(1)0.2990(1)1.0589(1)
RLSSVR-MCC()0.1770(3)0.1466(3)0.3019(3)0.9197(4)
RPSVR-MCC()0.1765(2)0.1461(2)0.2996(2)0.9185(5)