Walking Gait Phase Detection Based on Acceleration Signals Using Voting-Weighted Integrated Neural Network

Table 6

Summary of classification performance for different training functions.

Pace	Training function	Classification rate
		Accuracy (%)	Macro-F1 (%)	AUC
0.78 m/s	Bagging	76.9	70.7	0.98
	AdaBoosting	95.7	95.7	1.0
	VWI-DNN	99.6	99.5	1.0
1.0 m/s	Bagging	74.4	66.1	0.95
	AdaBoosting	92.4	91.9	1.0
	VWI-DNN	98.9	99.7	1.0
1.25 m/s	Bagging	74.4	66.3	0.98
	AdaBoosting	93.5	93.6	1.0
	VWI-DNN	99.1	99.7	1.0