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
|
#ifdef HAVE_CONFIG_H
# include "../config.h"
#endif
#include <stdio.h>
#include <stdlib.h>
#include <ghmm/matrix.h>
#include <ghmm/rng.h>
#include <ghmm/sequence.h>
#include <ghmm/model.h>
#include <ghmm/viterbi.h>
#include <ghmm/foba.h>
#include <ghmm/reestimate.h>
#include <ghmm/obsolete.h>
#include <ghmm/fbgibbs.h>
#include <ghmm/cfbgibbs.h>
#include <time.h>//time
int* getList(char* fileName, int count){
FILE *fp;
int c;
int* list = malloc(sizeof(int)*count);
if (!(fp = fopen(fileName, "rt"))) {
perror(fileName);
exit(1);
}
int i = 0;
while ((c = fgetc(fp)) != EOF && i < count ) {
if(c == 32){
c = 26;
}
else{
c -= 97;
}
//printf("%d ", c);
list[i] = c;
i++;
}
fclose(fp);
return list;
}
int main() {
int result;
/* Important! initialise rng */
ghmm_rng_init();
int count = 140400;
int* charList = getList("words.txt", count);
ghmm_dmodel my_model;
my_model.name = NULL;
my_model.model_type = GHMM_kDiscreteHMM;
my_model.silent = NULL;
my_model.maxorder = 0;
my_model.emission_history = 0;
my_model.tied_to = NULL;
my_model.order = NULL;
my_model.bp = NULL;
my_model.background_id = NULL;
my_model.topo_order =NULL;
my_model.topo_order_length = 0;
my_model.pow_lookup = NULL;
my_model.label = NULL;
my_model.label_alphabet = NULL;
my_model.alphabet = NULL;
ghmm_dstate model_states[2];
double symbols_vowel_state[27]={0.03906, 0.03537, 0.03537, 0.03909, 0.03583, 0.03630, 0.04048, 0.03537,0.03816, 0.03909, 0.03490, 0.03723, 0.03537, 0.03909, 0.03397, 0.03397, 0.03816, 0.03676, 0.04048, 0.03443, 0.03537, 0.03955, 0.03816, 0.03723, 0.03769, 0.03955, 0.03397};
double trans_prob_vowel_state[2]={0.47, 0.53};
double trans_prob_vowel_state_rev[2]={0.47, 0.53};
int trans_id_vowel_state[2]={0,1};
double symbols_consonant_state[27]={0.03732, 0.03408, 0.03455, 0.03828, 0.03782, 0.03922, 0.03688, 0.03408, 0.03875, 0.04062, 0.03735, 0.03968, 0.03548, 0.03735, 0.04062, 0.03595, 0.03641, 0.03408, 0.04062, 0.03548, 0.03922, 0.04062, 0.03455, 0.03595, 0.03408, 0.03408, 0.03688 };
double trans_prob_consonant_state[2]={0.51,0.49};
double trans_prob_consonant_state_rev[2]={0.51,0.49};
int trans_id_consonant_state[2]={0,1};
ghmm_dseq *my_output;
double log_p_viterbi, log_p_forward;
double **forward_alpha;
double forward_scale[count];
int *viterbi_path;
int i, pathlen;
/* flags indicating whether a state is silent */
int silent_array[2] = {0,0};
my_model.model_type = 0;
/* initialise vowel state */
model_states[0].pi = 0.49;
model_states[0].b=symbols_vowel_state;
model_states[0].out_states=2;
model_states[0].out_a=trans_prob_vowel_state;
model_states[0].out_id=trans_id_vowel_state;
model_states[0].in_states=2;
model_states[0].in_id=trans_id_vowel_state;
model_states[0].in_a=trans_prob_vowel_state_rev;
model_states[0].fix=0;
/* initialise consonant state */
model_states[1].pi = 0.51;
model_states[1].b=symbols_consonant_state;
model_states[1].out_states=2;
model_states[1].out_id=trans_id_consonant_state;
model_states[1].out_a=trans_prob_consonant_state;
model_states[1].in_states=2;
model_states[1].in_id=trans_id_consonant_state;
model_states[1].in_a=trans_prob_consonant_state_rev;
model_states[1].fix=0;
/* initialise model */
my_model.N=2;
my_model.M=27;
my_model.s=model_states;
my_model.prior=-1;
my_model.silent = silent_array;
fprintf(stdout,"transition matrix:\n");
ghmm_dmodel_A_print(stdout,&my_model,""," ","\n");
fprintf(stdout,"observation symbol matrix:\n");
ghmm_dmodel_B_print(stdout,&my_model,""," ","\n");
my_output = ghmm_dseq_calloc(2);
for(i=0;i< 2; i++){
my_output->seq[i] = charList;
my_output->seq_len[i] = count;
}
//ghmm_dseq_print(my_output, stdout);
//====================tests for fbgibbs==================================================
printf("fbgibbs \n");
clock_t t1, t2;//time
t1 = clock();//time
ghmm_dmodel* mo = ghmm_dmodel_copy(&my_model);
double **pA = NULL;
double **pB = NULL;
double *pPi = NULL;
init_priors(mo, &pA, &pB, &pPi);
int iter = 100;
int **Q = ghmm_dmodel_cfbgibbs(mo, my_output,
pA, pB, pPi,2, iter, 0);
printf("viterbi prob mcmc%f \n", ghmm_dmodel_viterbi_logp(mo, my_output->seq[0], my_output->seq_len[0], Q[0]));
printf("likelihood mcmc%f \n", ghmm_dmodel_likelihood(mo, my_output));
ghmm_dmodel_A_print(stdout,mo,""," ","\n");
ghmm_dmodel_B_print(stdout,mo,""," ","\n");
t2 = clock();//time
printf("time: %f\n", (double)(t2-t1)/CLOCKS_PER_SEC);
//=====================end test fbgibbs================================================
//=====================viterbi/em=====================================================
t1 = clock();
printf("Em/viterni\n\n");
//ghmm_dmodel_baum_welch_nstep(&my_model, my_output, 100, 0.0000001);
viterbi_path = ghmm_dmodel_viterbi(&my_model, my_output->seq[0],
my_output->seq_len[0],&pathlen, &log_p_viterbi);
//print
fprintf(stdout,
"(viterbi algorithm): %f\n",
log_p_viterbi);
printf("likelihood %f \n", ghmm_dmodel_likelihood(&my_model, my_output));
t2 = clock();//time
printf("time: %f\n", (double)(t2-t1)/CLOCKS_PER_SEC);
//==================================================================================
/* clean up */
//ghmm_dseq_free(&my_output);
free(viterbi_path);
ghmm_dmodel_free(&mo);
ighmm_cmatrix_free(&pA, my_model.N);
ighmm_cmatrix_free(&pB, my_model.N);
free(pPi);
return 0;
}
|
