libsent/src/phmm/gprune_beam.c
混合ガウス分布計算: Gaussian pruning (beam algorithm) [詳細]
#include <sent/stddefs.h>#include <sent/htk_hmm.h>#include <sent/htk_param.h>#include <sent/hmm.h>#include <sent/gprune.h>#include "globalvars.h"
関数 | |
| static void | clear_dimthres () |
| static void | set_dimthres () |
| static LOGPROB | compute_g_beam_updating (HTK_HMM_Dens *binfo) |
| Calculate probability with threshold update. | |
| static LOGPROB | compute_g_beam_pruning (HTK_HMM_Dens *binfo) |
| Calculate probability with pruning. | |
| boolean | gprune_beam_init () |
| void | gprune_beam_free () |
| void | gprune_beam (HTK_HMM_Dens **g, int gnum, int *last_id) |
| Compute a set of Gaussians with beam pruning. | |
変数 | |
| static LOGPROB * | dimthres |
| Threshold for each dimension (inversed). | |
| static int | dimthres_num |
| Length of above. | |
| static boolean * | mixcalced |
| Mark which Gaussian has been computed. | |
説明
混合ガウス分布計算: Gaussian pruning (beam algorithm)
- 日付:
- Thu Feb 17 03:27:53 2005
gprune_beam()は混合ガウス分布集合の計算ルーチンの一つです． beam pruning を使って上位のガウス分布の出力確率のみを高速に求めます． Tied-mixture HMM 使用時に Julius でGPRUNE_DEFAULT_BEAM が定義されているか， あるいはJuliusのオプション "-gprune beam" を指定することでこの関数が 使用されます．
beam pruning は最も積極的に枝刈りを行ないます．計算は最も高速ですが， 上位のガウス分布が正しく得られず出力確率の誤りが大きくなる可能性があります．
gprune_beam() は outprob_init() によってその関数へのポインタが compute_gaussset にセットされることで使用されます．このポインタが calc_tied_mix() または calc_mix() から呼び出されます．
- Revision
- 1.7
gprune_beam.c で定義されています。
関数
| static void clear_dimthres | ( | ) | [static] |
| static void set_dimthres | ( | ) | [static] |
Set beam thresholds by adding TMBEAMWIDTH to the maximum values of each dimension.
gprune_beam.c の 139 行で定義されています。
参照元 gprune_beam().
| static LOGPROB compute_g_beam_updating | ( | HTK_HMM_Dens * | binfo | ) | [static] |
Calculate probability with threshold update.
Calculate probability of a Gaussian toward OP_vec, while storing the maximum values of each dimension to dimthres. for future pruning. The pruning itself is not performed here. This function will be used to compute the first N Gaussians.
- 引数:
-
binfo [in] Gaussian density
- 戻り値:
- the output log probability.
gprune_beam.c の 158 行で定義されています。
参照元 gprune_beam().
| static LOGPROB compute_g_beam_pruning | ( | HTK_HMM_Dens * | binfo | ) | [static] |
Calculate probability with pruning.
Calculate probability of a Gaussian toward OP_vec, performing pruning using the dimension thresholds that has been set by compute_g_beam_updating() and set_dimthres().
- 引数:
-
binfo [in] Gaussian density
- 戻り値:
- the output log probability.
gprune_beam.c の 194 行で定義されています。
参照元 gprune_beam().
| boolean gprune_beam_init | ( | ) |
Initialize and setup work area for Gaussian pruning by beam algorithm.
- 戻り値:
- TRUE on success, FALSE on failure.
gprune_beam.c の 225 行で定義されています。
参照元 outprob_init().
| void gprune_beam_free | ( | ) |
| void gprune_beam | ( | HTK_HMM_Dens ** | g, | |
| int | gnum, | |||
| int * | last_id | |||
| ) |
Compute a set of Gaussians with beam pruning.
If the N-best mixtures in the previous frame is specified in last_id, They are first computed to set the thresholds for each dimension. After that, the rest of the Gaussians will be computed with those dimension thresholds to drop unpromising Gaussians from computation at early stage of likelihood computation. If the last_id is not specified (typically at the first frame of the input), a safe pruning as same as one in gprune_safe.c will be applied.
The calculated scores will be stored to OP_calced_score, with its corresponding mixture id to OP_calced_id. These are done by calling cache_push(). The number of calculated mixtures is also stored in OP_calced_num.
This can be called from calc_tied_mix() or calc_mix().
- 引数:
-
g [in] set of Gaussian densities to compute the output probability gnum [in] length of above last_id [in] ID list of N-best mixture in previous input frame, or NULL if not exist
gprune_beam.c の 277 行で定義されています。
参照元 outprob_init().
