GA与神经网络的结合的程序

frogfish

GA与神经网络的结合的程序

1. clear;
   
2. % Generate a set of points in the space
   
3. source_point = -10+20*rand(2,200);
   
4. 
   
5. % Classified boundary is y = sin(x)+2*cos^2(x)
   
6. points_c = zeros(1,200)
   
7. for i = 1:200
   
8.   point_c(i) = source_point(2,i)>(sin(source_point(1,i))+2*(cos(source_point(1,i))^2));
   
9. end
   
10. 
    
11. save training_data
    
12. % Plot the result
    
13. figure();
    
14. hold on;
    
15. for i = 1:200
    
16.   if point_c(i) == 1
    
17.       plot(source_point(1,i),source_point(2,i),'*');
    
18.   else
    
19.       plot(source_point(1,i),source_point(2,i),'+');
    
20.   end
    
21. end
    
22. %plot the boundary
    
23. x = -10:0.01:10;
    
24. y = sin(x)+2.*cos(x).^2;
    
25. plot(x,y,'r');
    
26. hold off;
    
27. 
    
28. % The training progress
    
29. % Defined the lower boundary
    
30. LB = zeros(1,25);
    
31. LB(1:18) = -1*ones(1,18);
    
32. LB(19:25) = -10*ones(1,7);
    
33. 
    
34. % Defined the upper boundary
    
35. UB = zeros(1,25);
    
36. UB(1:18) = 1*ones(1,18);
    
37. UB(19:25) = 10*ones(1,7);
    
38. 
    
39. % Set the training option
    
40. options=gaoptimset;
    
41. options.PopulationSize=20;
    
42. options.Generations=100;
    
43. options.StallGenLimit=Inf;
    
44. options.StallTimeLimit=Inf;
    
45. options = gaoptimset(options,'PlotFcns',{@gaplotbestf}')
    
46. [par,fval] = ga(@fitness_fuction,25,[],[],[],[],LB,UB,[],options);
    
47. 
    
48. % Generate 50 points to test the parameter
    
49. exam_point = -10+20*rand(2,50);
    
50. output = exam_function(par,exam_point);
    
51. 
    
52. % Plot the result
    
53. figure();
    
54. hold on;
    
55. for i = 1:50
    
56.   if output(i) == 1
    
57.       plot(exam_point(1,i),exam_point(2,i),'*');
    
58.   else
    
59.       plot(exam_point(1,i),exam_point(2,i),'+');
    
60.   end
    
61. end
    
62. %plot the boundary
    
63. x = -10:0.01:10;
    
64. y = sin(x)+2.*cos(x).^2;
    
65. plot(x,y,'r');
    
66. hold off;

复制代码

fitness function的源代码：

1. function fval = fitness_function(x);
   
2. 
   
3. load training_data;
   
4. 
   
5. h = zeros(6,200);
   
6. % Hidden layer output
   
7. h_out(1,:)= tanh(source_point(1,:)*x(1)+source_point(2,:)*x(2)+x(19));
   
8. h_out(2, :) = tanh(source_point(1,:)*x(3)+source_point(2,:)*x(4)+x(20));
   
9. h_out(3, :)= tanh(source_point(1,:)*x(5)+source_point(2,:)*x(6)+x(21));
   
10. h_out(4, :)= tanh(source_point(1,:)*x(7)+source_point(2,:)*x(8)+x(22));
    
11. h_out(5, :)= tanh(source_point(1,:)*x(9)+source_point(2,:)*x(10)+x(23));
    
12. h_out(6, :)= tanh(source_point(1,:)*x(11)+source_point(2,:)*x(12)+x(24));
    
13. 
    
14. % Output layer output
    
15. res = x(13:18)*h_out+x(25);
    
16. zm = zeros(1,200);
    
17. om = ones(1,200);
    
18. res(reszm) = om(res>zm);
    
19. fval = sum(abs(point_c-res));

复制代码

检测程序的源代码：

1. function res = exam_function(x,data);
   
2. h = zeros(6,50);
   
3. % Hidden layer output
   
4. h_out(1, :)= tanh(data(1,:)*x(1)+data(2,:)*x(2)+x(19));
   
5. h_out(2, :)= tanh(data(1,:)*x(3)+data(2,:)*x(4)+x(20));
   
6. h_out(3, :)= tanh(data(1,:)*x(5)+data(2,:)*x(6)+x(21));
   
7. h_out(4, :)= tanh(data(1,:)*x(7)+data(2,:)*x(8)+x(22));
   
8. h_out(5, :)= tanh(data(1,:)*x(9)+data(2,:)*x(10)+x(23));
   
9. h_out(6, :)= tanh(data(1,:)*x(11)+data(2,:)*x(12)+x(24));
   
10. 
    
11. % Output layer output
    
12. res = x(13:18)*h_out+x(25);
    
13. zm = zeros(1,50);
    
14. om = ones(1,50);
    
15. res(reszm) = om(res>zm);

复制代码

zwf100

有点意思哦