train.cc File Reference

This file contains the program to estimate Hidden Markov Model parameter using training sequences. More...

Include dependency graph for train.cc:

Go to the source code of this file.

Functions
int	main (int argc, char *argv[])
success_t	train_baumwelch ()
success_t	train_baumwelch_discrete ()
success_t	train_baumwelch_gaussian ()
success_t	train_baumwelch_mixture ()
success_t	train_viterbi ()
success_t	train_viterbi_discrete ()
success_t	train_viterbi_gaussian ()
success_t	train_viterbi_mixture ()
void	usage ()
Variables
const fx_module_doc	hmm_train_main_doc
const fx_entry_doc	hmm_train_main_entries []
const fx_submodule_doc	hmm_train_main_submodules []

---

Detailed Description

This file contains the program to estimate Hidden Markov Model parameter using training sequences.

It use two algorithm: Baum-Welch (EM) and Viterbi

Usage: train --type=TYPE --profile=PROFILE --seqfile=FILE [OPTIONS] See the usage() function for complete option list

Definition in file train.cc.

---

Variable Documentation

const fx_module_doc hmm_train_main_doc

Initial value:

 {
  hmm_train_main_entries, hmm_train_main_submodules,
  "This is a program training HMM models from data sequences. \n"
}

Definition at line 82 of file train.cc.

const fx_entry_doc hmm_train_main_entries[]

Initial value:

 {
  {"type", FX_REQUIRED, FX_STR, NULL,
   "  HMM type : discrete | gaussian | mixture.\n"},
  {"algorithm", FX_PARAM, FX_STR, NULL,
   "  Training algoritm: baumwelch | viterbi.\n"},
  {"seqfile", FX_REQUIRED, FX_STR, NULL,
   "  Output file for the data sequences.\n"},
  {"guess", FX_PARAM, FX_STR, NULL,
   "  File containing guessing HMM model profile.\n"},
  {"numstate", FX_PARAM, FX_INT, NULL,
   "  If no guessing profile specified, at least provide the number of states.\n"},
  {"profile", FX_REQUIRED, FX_STR, NULL,
   "  Output file containing trained HMM profile.\n"},
  {"maxiter", FX_PARAM, FX_INT, NULL,
   "  Maximum number of iterations, default = 500.\n"},
  {"tolerance", FX_PARAM, FX_DOUBLE, NULL,
   "  Error tolerance on log-likelihood as a stopping criteria.\n"},
  FX_ENTRY_DOC_DONE
}

Definition at line 58 of file train.cc.

const fx_submodule_doc hmm_train_main_submodules[]

Initial value:

 {
  FX_SUBMODULE_DOC_DONE
}

Definition at line 78 of file train.cc.