HMM.gz_HMM 状态_隐马尔可夫模型

#include<stdlib.h> #include<stdio.h> #include"HMM.h" HMM::HMM(int size,int strmaxlen) { states_size=size; maxlen=strmaxlen; states_transform=(long double *)malloc((states_size*states_size)*sizeof(long double)); words_subject=(long double *)malloc((states_size*26)*sizeof(long double)); sta_it_poss=(long double *)malloc((strmaxlen*states_size)*sizeof(long double)); states_start=(long double *)malloc(states_size*sizeof(long double)); best_sta_squ_poss=(long double *)malloc((strmaxlen*states_size)*sizeof(long double)); best_sta_squ=(int *)malloc((strmaxlen*states_size)*sizeof(int)); best_squ=(int *)malloc(strmaxlen*sizeof(int)); words=(char *)malloc(strmaxlen*sizeof(char)); for(int j=0;j<states_size*strmaxlen;j++)sta_it_poss[j]=0; } bool HMM::Set_states_transform(long double*transform) { for(int i=0;i<states_size*states_size;i++) states_transform[i]=transform[i]; return 0; } bool HMM::Set_words_subject(long double *subject) { for(int i=0;i<states_size*26;i++) words_subject[i]=subject[i]; return 0; } bool HMM::Set_states_start(long double*start) { for(int i=0;i<states_size;i++) states_start[i]=start[i]; return 0; } bool HMM::Set_words(char *str) { int len=0; for(len=0;str[len]!='\0';len++) words[len]=str[len]; words[len]='\0'; } bool HMM::Init_sta_it_poss() { for(int i=0;i<maxlen*states_size;i++)sta_it_poss[i]=0; for(int j=0;j<states_size;j++) sta_it_poss[j]=states_start[j]*words_subject[j*26+words[0]-'a'];//初始化时刻0,在状态j的概率 } long double HMM::forward(int i,int t) { int h=0; if(words==NULL)return 0; if(words[i]=='\0') { return sta_it_poss[(i-1)*states_size+t]; } if(i==0)return sta_it_poss[t]; else { for(h=0;h<states_size;h++) { sta_it_poss[t+states_size*i]+=sta_it_poss[h+states_size*(i-1)]*states_transform[h*states_size+t]*words_subject[t*26+words[i]-'a']; } } } long double HMM::backward(int j,int t) { int h=0;//h,t表示状态,j表示时刻 if(words==NULL||j<0)return 0; if(j==0)return sta_it_poss[t]; if(j>0) { for(h=0;h<states_size;h++) { if(sta_it_poss[(j-1)*states_size+h]==0)sta_it_poss[(j-1)*states_size+h]=backward(j-1,h); } for(h=0;h<states_size;h++) { sta_it_poss[j*states_size+t]+=sta_it_poss[(j-1)*states_size+h]*states_transform[h*states_size+t]*words_subject[t*26+words[j]-'a']; } return sta_it_poss[j*states_size+t]; } } bool HMM::Init_best_sta_squ() { for(int i=0;i<maxlen*states_size;i++)best_sta_squ_poss[i]=0; for(int j=0;j<states_size;j++) { best_sta_squ_poss[j]=states_start[j]*words_subject[j*26+words[0]-'a']; best_sta_squ[j]=j; } } long double HMM::viterbi(int i,int t) { long double temp=0; if(i==0) { return best_sta_squ_poss[t]; } if(i>0) { for(int h=0;h<states_size;h++) { if(best_sta_squ_poss[(i-1)*states_size+h]==0) best_sta_squ_poss[(i-1)*states_size+h]=viterbi(i-1,h); temp=best_sta_squ_poss[(i-1)*states_size+h]*states_transform[h*states_size+t]*words_subject[t*26+words[i]-'a']; if(temp>=best_sta_squ_poss[i*states_size+t]) {best_sta_squ_poss[i*states_size+t]=temp;best_sta_squ[i*states_size+t]=h;} } return best_sta_squ_poss[i*states_size+t]; } } long double HMM::Get_best_sta_squ_poss() {int len=0,t=0; long double temp=0; for(len=0;words[len]!='\0';len++); for(t=0;t<states_size;t++) viterbi(len-1,t); for(t=0;t<states_size;t++){ if(temp<best_sta_squ_poss[(len-1)*states_size+t]){temp=best_sta_squ_poss[(len-1)*states_size+t];best_squ[len-1]=t;} best_sta_squ[(len-1)*states_size+t]=best_squ[len-1];} return temp; } bool HMM::Get_best_squ() { int len=0,i=0; for(len=0;words[len]!='\0';len++); for(i=len-2;i>=0;i--) best_squ[i]=best_sta_squ[(i+1)*states_size+best_squ[i+1]]; for(i=0;i<len;i++) printf("%d ",best_squ[i]); printf("\n"); }