-
Notifications
You must be signed in to change notification settings - Fork 7
/
RBFnetwork.m
176 lines (94 loc) · 4.79 KB
/
RBFnetwork.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
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
classdef RBFnetwork
properties
neural;
gamma;
size;
layers = 3;
end
methods
function obj = RBFnetwork(net, rate, data)
obj.gamma = rate;
obj.size = net;
for i = 1 : 3
for j = 1 : obj.size(i)
if i == 1
layer(j) = Neuron(true);
elseif i == 2
layerR(j) = RBFneuron(data(j, :));
else
layer(j) = Neuron(net(i - 1));
end
end
if i ~= 2
obj.neural{i} = {layer(1 : net(i))};
else
obj.neural{i} = {layerR(1 : net(i))};
end
end
end
function obj = UncontrolledLearning(obj, dataset, epsilon)
delta = 5;
len = length(dataset);
while abs(delta) >= epsilon
delta = 0;
for i = 1 : len
obj = Run(obj, dataset(i, :));
index = Minimal(obj);
[obj.neural{2}{1}(index), delt] = WeightsChange(obj.neural{2}{1}(index), obj.gamma, dataset(i, :));
delta = delta + abs(delt);
end
delta = delta / len
end
beta = 0.3;
for i = 1 : obj.size(2)
mini = 0;
for j = 1 : obj.size(2)
if j ~= i
obj.neural{2}{1}(j) = Activate(obj.neural{2}{1}(j), obj.neural{2}{1}(i).weights(1 : end - 1));
if mini == 0
mini = obj.neural{2}{1}(j).input;
elseif obj.neural{2}{1}(j).input < mini
mini = obj.neural{2}{1}(j).input;
end
end
end
obj.neural{2}{1}(i).weights(end) = mini / beta;
end
end
function obj = Learning(obj, data, answers)
obj = Run(obj, data);
for i = 1 : obj.size(3)
obj.neural{3}{1}(i).error = answers(i) - obj.neural{3}{1}(i).input;
obj.neural{3}{1}(i).err = abs(obj.neural{3}{1}(i).error);
for j = 1 : obj.size(2)
obj.neural{3}{1}(i).weights(j) = obj.neural{3}{1}(i).weights(j) + obj.gamma * obj.neural{3}{1}(i).error * obj.neural{2}{1}(j).output;
end
obj.neural{3}{1}(i).weights(obj.size(2) + 1) = obj.neural{3}{1}(i).weights(obj.size(2) + 1) - obj.gamma * obj.neural{3}{1}(i).error;
end
end
function r = Minimal(obj)
mini = obj.neural{2}{1}(1).input;
r = 1;
for i = 2 : obj.size(2)
if mini > obj.neural{2}{1}(i).input
mini = obj.neural{2}{1}(i).input;
r = i;
end
end
end
function obj = Run(obj, data)
for i = 1 : obj.size(1)
obj.neural{1}{1}(i).output = data(i);
end
datas = zeros(1, max(obj.size));
for i = 2 : 3
for j = 1 : obj.size(i)
for k = 1 : obj.size(i - 1)
datas(k) = obj.neural{i - 1}{1}(k).output;
end
obj.neural{i}{1}(j) = Activate(obj.neural{i}{1}(j), datas(1 : obj.size(i - 1)));
end
end
end
end
end