Gait Recognition for Human-Exoskeleton System in Locomotion Based on Ensemble Empirical Mode Decomposition

Table 2

Gait recognition accuracy for gait recognition mode A based on the IMFs of the original signal for the 14 subjects who walked without exoskeletons (mean  SD%).

Component	SVM	Kmeans	Decision tree	Logistic regression	Naive Bayes	Random forest
IMF1	45.10  7.49	36.13  6.97	36.62  5.63	44.63  7.69	34.23  6.20	38.33  6.09
IMF2	45.11  7.61	41.64  5.53	42.00  5.24	45.20  7.74	35.55  6.79	44.36  5.81
IMF3	45.65  7.58	74.32  9.56	67.86  9.90	46.35  7.03	53.79  11.74	72.47  9.47
IMF4	92.45  9.22	98.13  1.26	95.17  2.65	76.79  10.41	90.65  5.34	98.08  1.69
IMF5	98.59  0.98	98.62  0.95	97.75  1.85	97.96  2.75	91.20  4.96	98.30  1.24
IMF6	98.64  1.00	98.45  1.08	98.05  1.58	98.29  1.39	91.36  4.17	98.32  1.36
IMF7	93.52  8.11	98.42  1.17	98.22  1.56	88.90  5.94	84.67  9.65	98.47  1.29
IMF8	98.20  1.14	98.59  1.05	98.45  1.35	80.94  16.84	83.31  10.76	98.56  1.24