Classification of Electroencephalography Based on BP Neural Network Optimized By Crossover Operation of Artificial Bee Colonies

Trialfunction	Functionname	Function expression	Search scope	Optimalvalue
Ⅰ	Griewank	$f_{5} (r) = \frac{1}{4000} \overset{Q}{\sum_{i = 1}} r_{i}^{2} - \overset{Q}{\prod_{i = 1}} \cos (\frac{r_{i}}{\sqrt[]{i}}) + 1$	[-10,10]	f₅(0)=0
Ⅱ	Rastrigin	$f_{2} (r) = \overset{Q}{\sum_{i = 1}} [r_{i}^{2} - 10 \cos (2 π r_{i}) + 10]$	[-5.12,5.12]	f₂(0)=0
Ⅲ	Sphere	$f_{1} (r) = \overset{Q}{\sum_{i = 1}} r_{i}^{2}$	[-5.12,5.12]	f₁(0)=0
Ⅳ	Rosenbrock	$f_{6} (r) = \overset{Q - 1}{\sum_{i = 1}} [100 (r_{i + 1} - r_{i}^{2})^{2} + (r_{i} {- 1)}^{2}]$	[-2.048,2.048]	f₆(0)=0

Table 2. Performance comparison of three algorithms

View table

Table 2. Performance comparison of three algorithms

Trial function	Method	Optimal value	Convergence rate /s
Ⅰ	SABC	3.83×10^-5	900.9	3.88×10^-3	1.04×10^-16
	GABC	0	230.0	8.30×10^-13	1.41×10^-17
	CGABC	0	191.4	5.47×10^-16	0
Ⅱ	SABC	0	122.1	0	2.23×10^-14
	GABC	0	23.2	0	0
	CGABC	0	15.2	0	0
Ⅲ	SABC	1.05×10^-15	28.7	5.20×10^-15	9.38×10^-17
	GABC	3.10×10^-18	17.2	8.03×10^-13	5.08×10^-17
	CGABC	9.43×10^-18	14.3	4.31×10^-18	2.66×10^-17
Ⅳ	SABC	3.10×10^-15	752.3	4.20×10^-15	0.98
	GABC	9.10×10^-16	22.3	5.45×10^-15	0.61
	CGABC	1.03×10^-18	19.1	8.76×10^-17	0.11

Table 3. Training results of different BP neural networks
View table
Table 3. Training results of different BP neural networks
Neural network Max iterations Min interations Average interations Target error
BP 13465 7732 9641 10^-4
GABC-BP 321 164 215 10^-4
CGABC-BP 214 61 175 10^-4

Table 4. Accuracy results of feature classification of different BP neural networks unit:%
View table
Table 4. Accuracy results of feature classification of different BP neural networks unit:%
Neural network P1 P2 P3 P4 P5 P6 P7 Average accuracy
BP 81.2 84.5 83.7 85.7 82.4 83.2 82.9 83.4
GABC-BP 86.3 85.4 88.1 87.3 86.8 87.4 84.7 86.6
CGABC-BP 90.7 89.5 93.9 92.0 91.1 90.6 92.5 91.5

Tools

Get Citation

Copy Citation Text

Xu Jian, Chen Qianqian, Liu Xiuping. Classification of Electroencephalography Based on BP Neural Network Optimized By Crossover Operation of Artificial Bee Colonies[J]. Laser & Optoelectronics Progress, 2020, 57(21): 212001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites