-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathGasNN.m
69 lines (65 loc) · 1.37 KB
/
GasNN.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
function NGNnetwork = GasNN(X, params)
%% Load
nData = size(X,1);
nDim = size(X,2);
X = X(randperm(nData),:);
Xmin = min(X);
Xmax = max(X);
%% Params
N = params.N;
MaxIt = params.MaxIt;
tmax = params.tmax;
epsilon_initial = params.epsilon_initial;
epsilon_final = params.epsilon_final;
lambda_initial = params.lambda_initial;
lambda_final = params.lambda_final;
T_initial = params.T_initial;
T_final = params.T_final;
%% Initialization
w = zeros(N, nDim);
for i = 1:N
w(i,:) = unifrnd(Xmin, Xmax);
end
C = zeros(N, N);
t = zeros(N, N);
tt = 0;
%% Body
for it = 1:MaxIt
for l = 1:nData
% Slect Input Vector
x = X(l,:);
% Competion and Ranking
d = pdist2(x,w);
[~, SortOrder] = sort(d);
% Calculate Parameters
epsilon = epsilon_initial*(epsilon_final/epsilon_initial)^(tt/tmax);
lambda = lambda_initial*(lambda_final/lambda_initial)^(tt/tmax);
T = T_initial*(T_final/T_initial)^(tt/tmax);
% Adaptation
for ki = 0:N-1
i=SortOrder(ki+1);
w(i,:) = w(i,:) + epsilon*exp(-ki/lambda)*(x-w(i,:));
end
tt = tt + 1;
% Creating Links
i = SortOrder(1);
j = SortOrder(2);
C(i,j) = 1;
C(j,i) = 1;
t(i,j) = 0;
t(j,i) = 0;
% Aging
t(i,:) = t(i,:) + 1;
t(:,i) = t(:,i) + 1;
% Remove Old Links
L = t(i,:)>T;
C(i, L) = 0;
C(L, i) = 0;
end
% disp(['Iteration ' num2str(it)]);
end
%% Results
NGNnetwork.w = w;
NGNnetwork.C = C;
NGNnetwork.t = t;
end