Biomedical Engineering Reference
In-Depth Information
produces the output. Only with a value of itera-
tions greater than 1 the time decreased activation
option is active), Activation Function and Neurons
(number of hidden neurons).
We test with different combinations of GA
parameters and the best one is the following:
As we can see in the Figure 12, a time decreased
activation of 0.3 shows the best evolution for 10
internal iterations in the ANN.
Table 2 shows the configuration of the ANN
parameters in order to test the system and the
simulated time decreased activation with 5 in-
ternal iterations in the ANN. We also stop the
genetic algorithm after 200 generations.
We can see the evolution of the Genetic Al-
gorithm in the Figure 13.
As we can see in the Figure 13, in this case, a
time decreased activation of 0.6 shows the best
evolution for 5 internal iterations in the ANN. In
this case the RANN needs more speed to make
the simulation of the action potential between
cycles.
Table 3 shows the configuration of the ANN
parameters in order to test the system and the
simulated time decreased activation with 10 inter-
nal iterations in the ANN and Sigmoid activation
function. We again stop the genetic algorithm
after 200 generations.
As we can see in the Figure 14, a time decreased
activation of 0.01 shows the best evolution for 10
internal iterations in the ANN.
Individuals:
200
Crossover Rate:
90%
Mutation Rate:
10%
Selection operation:
Montecarlo
Substitute Strategy:
Substitute Worst
Cross Points:
1 Cross Point
Results for Number of Sunspots
Time Series
Table 1 shows the configuration of the ANN
parameters in order to test the system and the
simulated time decreased activation with 10
internal iterations in the ANN and Hyperbolic
activation function. We stop the genetic algorithm
after 200 generations.
We can see the evolution of the Genetic Al-
gorithm in the Figure 12.
Table 3. Sunspots ANN parameters
Table 1. Sunspots ANN parameters
Mean
Square
Error
Mean
Square
Error
Time
Iterations
Act. Funct.
Neurons
Time
Iterations
Act. Funct.
Neurons
0.01 10
Hyperbolic
1
0.012
0.01 10
Sigmoid
1
0.013
0.1
10
Hyperbolic
1
0.010
0.1
10
Sigmoid
1
0.017
0.3
10
Hyperbolic
1
0.010
0.3
10
Sigmoid
1
0.018
0.6
10
Hyperbolic
1
0.010
0.6
10
Sigmoid
1
0.016
--
1
Hyperbolic
1
0.019
--
1
Sigmoid
1
0.024
Table 4. Sunspots ANN parameters
Table 2. Sunspots ANN parameters
Mean
Square
Error
Mean
Square
Error
Time
Iterations
Act. Funct.
Neurons
Time
Iterations
Act. Funct.
Neurons
0.01 5
Sigmoid
1
0.019
0.01 5
Hyperbolic
1
0.012
0.1
5
Sigmoid
1
0.017
0.1
5
Hyperbolic
1
0.013
0.3
5
Hyperbolic
1
0.010
0.3
5
Sigmoid
1
0.020
0.6
5
Hyperbolic
1
0.010
0.6
5
Sigmoid
1
0.018
--
1
Hyperbolic
1
0.019
--
1
Sigmoid
1
0.024
Search WWH ::
Custom Search