-
Notifications
You must be signed in to change notification settings - Fork 0
/
SOMClassifier (2).cpp
278 lines (223 loc) · 5.9 KB
/
SOMClassifier (2).cpp
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
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
// SOMClassifier.cpp : Defines the entry point for the console application.
//
#include <fstream>
#include <stdio.h>
#include <tchar.h>
#include <iostream>
#include <math.h>
#include <vector>
#include <fstream>
#include <string>
#include <sstream>
#include "EasyBMP\EasyBMP.h"
#include <time.h>
#include <math.h>
#include "stdafx.h"
#define NUM_NEURONS 1000
struct Neuron{
int X;
int Y;
int Z;
};
Neuron * d_neurons;
Neuron * neurons;
float * weights;
float * d_weights;
float * input_data;
float * d_input_data;
using namespace std;
int numIterations = 1000;
float initLearningRate = .9;
int mapWidth = 10, mapHeight = 10, mapDepth = 10;
int initialMapRadius = 5;
string int_to_str(int i){
stringstream ss;
ss << i;
string str = ss.str();
return str;
}
float map(float value, float istart, float istop, float ostart, float ostop) {
return ostart + (ostop - ostart) * ((value - istart) / (istop - istart));
}
void readData(float * input_data){
int numSamples = 40;
string subject_id = "";
string face_position_num = "";
string face_num = "";
string final = "";
BMP image;
int count = 0;
int countImage = 0;
for(int i = 1; i<=numSamples; i++){
subject_id = "s"+int_to_str(i);
for(int j = 1; j<=10; j++){
face_position_num = int_to_str(j);
for(int k = 0; k<3; k++){
face_num = int_to_str(k);
final = "..\\images\\"+subject_id+"\\"+face_position_num+"\\"+face_num+".bmp";
image.ReadFromFile(final.c_str());
//face_position is the orientation ID of the face
for(int l = 0; l< 23; l++){
for(int m = 0; m<28; m++){
input_data[countImage] = map((((float)image(l, m)->Red)/255), 0, 1, -1, 1);
countImage++;
}
}
}
}
}
}
double calcDistBetweenNodes(Neuron n1, Neuron n2){
double temp = (double)((n1.X-n2.X)*(n1.X-n2.X)+(n1.Y-n2.Y)*(n1.Y-n2.Y)+(n1.Z-n2.Z)*(n1.Z-n2.Z));
return sqrt(temp);
}
int findBMU(float * inputVector, float * weights){
int count = 0;
float currentDistance = 0;
int winner = 0;
float leastDistance = 99999;
//if(i<10&&j<10&&k<10){
for(int i = 0; i<10; i++){
for(int j = 0;j<10; j++){
for(int k = 0; k<10; k++){
int offset = (i*100+j*10+k)*644;
for(int i = offset; i<offset+644; i++){
currentDistance = (inputVector[count]-weights[i])*(inputVector[count]-weights[i]);
count++;
}
count = 0;
if(currentDistance<leastDistance){
winner = offset;
leastDistance = currentDistance;
}
}
//}
}
}
return winner;
}
float mapRadius(int time){
double initialMapRadius = max(mapWidth, mapHeight)/1.5;
double timeConstant = numIterations/log(initialMapRadius);
double radius = initialMapRadius*exp(-(time/timeConstant));
return radius;
}
double learningRate(int time){
float iterations = (float)numIterations;
double rate = initLearningRate*exp((-(time/iterations)));
return rate;
}
double theta(float distanceBetweenNodes, float radius){
return exp(-(distanceBetweenNodes*distanceBetweenNodes)/(2*radius*radius));
}
void train(float *weights, Neuron*neurons, float*input_data){
Neuron winningNeuron;
int winningNeuronID = 0;
int subjectNum;
int positionNum;
int winX;
int winY;
int winZ;
float neighboorhoodRadius;
float rate;
float distance;
double coeff;
for(int y = 0; y<numIterations; y++){
//select a random image
positionNum = rand()%10;
subjectNum = rand()%4;
float * data = new float[644];
int count = 0;
for(int i = (rand()%1200)*644; i<(rand()%1200)*644+644; i++){
data[count] = input_data[i];
count++;
}
findBMU(data, weights);
winningNeuronID = winningNeuronID/644;
neighboorhoodRadius = mapRadius(y);
rate = learningRate(y);
winX = (winningNeuronID/644)/100;
winY = ((winningNeuronID/644)-(winX*100))*10;
winZ = ((winningNeuronID/644)-(winX*100)-(winY*10));
printf("1...");
for(int h = 0; h<mapWidth; h++){
for(int i = 0; i<mapHeight; i++){
for(int j = 0; j<mapDepth; j++){
distance = calcDistBetweenNodes(neurons[h*100+i*10+j], neurons[winningNeuronID/644]);
if(distance<neighboorhoodRadius){
coeff = theta(distance, neighboorhoodRadius)*rate;
float * newWeight;
newWeight = new float [644];
for (int w = 0; w<644; w++){
double diff = weights[(h*100+i*10+j)*644+w];
newWeight[w] =diff*coeff;
}
for (int w = 0; w<644; w++){
weights[(h*100+i*10+j)*644+w]+=newWeight[w];
}
delete newWeight;
}
}
}
}
printf("2 \n");
if(y%10 == 0){
cout << y << endl;
}
}
}
void setXYZ(Neuron * neurons){
for(int i = 0; i<10; i++){
for(int j = 0; j<10; j++){
for(int k = 0; k<10; k++){
neurons[i*100+j*10+k].X = i;
neurons[i*100+j*10+k].Y = j;
neurons[i*100+j*10+k].Z = k;
}
}
}
}
//this is done on HOST side because rand() can't be called device side
void setWeights(float * weights){
for(int i = 0; i<NUM_NEURONS*644; i++){
weights[i] = (double)rand()/RAND_MAX;
}
}
int main(int argc, char*argv[])
{
printf("asdf");
neurons = new Neuron[1000];
printf("asdf");
setXYZ(neurons);
printf("Stage 1 complete \n");
srand(time(NULL));
weights = new float[NUM_NEURONS*644];
setWeights(weights);
printf("Stage 2 complete \n");
input_data = new float[1200*644];
readData(input_data); // read data with host array
printf("Training started \n");
train(weights, neurons, input_data);
}
/*
float * test_vector;
test_vector = (float*)malloc(644*sizeof(float));
for(int i = 0; i<644; i++){
test_vector[i] = .32f;
}
float * d_test_vector;
cudaMalloc(&d_test_vector, 644*sizeof(float));
cudaMemcpy(d_test_vector, test_vector, 644*sizeof(float), cudaMemcpyHostToDevice);
float * d_least;
float * least;
least = (float*)malloc(sizeof(float));
*least = 9999999;
cudaMalloc(&d_least, sizeof(float));
cudaMemcpy(d_least, least, sizeof(float), cudaMemcpyHostToDevice);
int * d_winner;
int * winner;
winner = (int*)malloc(sizeof(int));
*winner = 0;
cudaMalloc(&d_winner, sizeof(int));
cudaMemcpy(d_winner, winner, sizeof(int), cudaMemcpyHostToDevice);
*/