| 基于MMD算法的HMM参数重估计的矩阵表示 |
| |
| 引用本文： | 张道鸿,李兵.基于MMD算法的HMM参数重估计的矩阵表示[J].数学理论与应用,2005,25(3):19-23. |
| |
| 作者姓名： | 张道鸿 李兵 |
| |
| 作者单位： | [1]国防科学技术大学理学院,长沙 410073 |
| |
| 摘 要： | 本文提出的MMD算法用于提高模型区别错误信息和正确信息的能力．利用该算法在对模型的参数进行重估计时．涉及到复杂的目标函数的梯度运算．击运用矩阵运算使得梯度运算变得简单明了，因此本文给出了MMD算法下的HMM参数重估计的矩阵表示形式并给出了证明．
|
| 关 键 词： | 最大模型距离 梯度法 隐马尔可夫模型 重估计 矩阵表示 D算法 HMM 估计 矩阵运算 错误信息 |
| 收稿时间： | 04 13 2005 12:00AM |
Matrix Expression of HMM Parameters Re-estimation Based on MMD Algorithm |
| |
| Zhang Daohong, Li Bing ,.Matrix Expression of HMM Parameters Re-estimation Based on MMD Algorithm[J].Mathematical Theory and Applications,2005,25(3):19-23. |
| |
| Authors: | Zhang Daohong Li Bing |
| |
| Institution: | College of Science,National Univ. of Defense Technology,Changsha,410073 |
| |
| Abstract: | This paper presents the MMD algorithm to improve the ability of HMM of differentiating error and rig information.When we use this algorithm to re-estimate the parameters of HMM,complex gradient operations are involve in.However ,matrix operation can make gradient operations become simpleness.So we gained the matrix expression of HMM parameters re-estimation under MMD and proofed it at last. |
| |
| Keywords: | maximum model distance gradient method hidden Markov models re--estimation |
| 本文献已被 CNKI 维普 等数据库收录！ |
