Scribd
Upload
4 views2 pages

Algorithm Design Sample

The document outlines an algorithm for Independent Component Analysis (ICA) using Particle Swarm Optimization (PSO). It details the initialization of variables, the iterative process for updating particle positions and velocities, and the calculation of fitness values to recover sources from whitened vectors. The algorithm continues until a specified number of iterations is reached, resulting in the separated vectors as output.

Uploaded by

ponasi2280
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
4 views2 pages

Algorithm Design Sample

The document outlines an algorithm for Independent Component Analysis (ICA) using Particle Swarm Optimization (PSO). It details the initialization of variables, the iterative process for updating particle positions and velocities, and the calculation of fitness values to recover sources from whitened vectors. The algorithm continues until a specified number of iterations is reached, resulting in the separated vectors as output.

Uploaded by

ponasi2280
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Algorithm (3.

5): ICA based on PSO Algorithm


Input: white-x % whitened vectors
Output: z; % separated vectors (recovered sources)

% Variables Definition
K: number of sources % integer variable
population: population size % integer variable (20 – 100)
vi : velocity of the particles % real variable
xi: position of the particles % real variable
r1,r2: two uniform random numbers % real variables [0-1]
c1,c2: two acceleration coefficients % fixed variables ( ≥0 )
vmax: maximum velocity value of current particle
-vmax: minimum velocity value of current particle
pgmax: maximum global fitness value of particle % real variable
pimax: maximum local fitness value of particle % real variable
fitmax: list of maximum local fitness values % real vector
fgmax: list of maximum global fitness values % real vector
fit: initial fitness values of current positions of particle % real vector.
fitnew: list of next fitness values % real vector
iteration: maximum iteration % fixed variables ( ≥ 8)
Begin
1. Initializing set of diagonal separated matrices.
2. x=rand( K, K, population);
3. Initializing the velocities of the particles
4. v = rand( K, K, population) * vmax; % real vector
5. Calculate initially fitness values of the current positions of particles the objective function
6. For i=1 to population
7. y= x* white-x;
8. Centering and Whitening y;
9. The proposed objective function
10. fit(i) = sum(proposed fun.); % compute current fitness value
11. End_for
12. Set the initialize values of the algorithm parameters.
13. Main loop iteration of the algorithm for each particle.
14. d=1; % iteration index
15. Do
16. for i=1 to population
17. for j=1 to K
18. for s=1 to K
19. vj,s,i=vj,s,i+c1*r1* (pimaxj,s,i – xj,s,i )+c2*r1*(pgmaxj,s - xj,s,i);
20. % Evaluate vj,s,i with the vmax and -vmax parameters
21. xj,s,i= xj,s,i + vj,s,i ;
22. end_for(s)
23. end_for(j)
24. end_for(i)
25. Calculate new fitness values of the positions of particles
26. For m=1 to population
27. y= white-x * x;
28. Centering and Whitening y;
29. The proposed objective function
30. fitnew(m)=sum(proposed fun.); % compute new fitness value
31. End_for(m)
32. Set maximum values of fitnew and pgmax as maximum fitness values and its positions respectively.
33. Get the maximum values of the fitness values.
34. d=d+1; % increment the iteration counter.
35. Until d= iteration; % terminate the loop of PSO algorithm.
36. z = y; % recovered sources
End Algorithm.

You might also like