classifiers/nb.h

Go to the documentation of this file.

/* Author: Andrew I. Schein    -*- c++ -*- */

#ifndef NB_H
#define NB_H

#include "../hash/nps_vocab.h"
#include "../imat/imat_csr.h"
#include "../error/error.h"

/* nb_counts_t is for tracking counts. When counting is complete, transform
   into a model (separate struct) */
   
typedef struct {
  long nCat;
  long nFeat; 
  double *restrict cat;      /* [cat] count */
  double *restrict cat_feat; /* [cat][feat] count */
} nb_counts_t;

extern nb_counts_t *nb_counts_new(long,long);
extern void nb_counts_del(nb_counts_t *);

/* nb_add_counts: accumulate instance into the counts */
extern void nb_add_counts(nb_counts_t *, nps_imat_csr_t *); 

/* smoothing strategies.  Just Laplace, for now. */
enum nb_smooth_t { Laplace=0  };

/* nb_model_t: the trained model */
typedef struct {
  long nCat;
  long nFeat;
  bio_t *cat;      /* -log p(c) */
  bio_t *cat_feat; /* - log p(w|c), but with opposite indexing */
} nb_model_t;

extern long nb_model_new(long,long);
extern void nb_model_del(nb_model_t *);
extern nb_model_t *nb_ceate_model(nb_counts_t *,int); /* turn cont data into model */
extern long nb_classify(nb_model_t *, char, s8_t *, double *, long, double *);

extern int nb_model_save(nb_model_t *, NPS_VOC_T * , NPS_VOC_T * , NPS_VOC_T * ,const char*, nps_error_t *err);
extern nb_model_t* nb_model_load(nb_model_t **, NPS_VOC_T **, NPS_VOC_T ** ,
                                 const char *, 
                                 char ignore_marginals, 
                                 nps_error_t *err);

#endif

---