Baum-Welchアルゴリズムの動作と応用例

← Back to document page

Baum-Welch . Step by Step the Baum-Welch Algorithm and its Application Jin'ichi MURAKAMI. . EM EM. . EM Baum-Welch . Baum-Welch . Baum-Welch . Baum-Welch , EM . Baum-Welch . 1. . . EM 2. HMM. EM (Expectation- Hidden Markov Model . maximization algorithm) 1 3. HMM . EM . 4. HMM . 5. HMM . Baum-Welch . 6. Baum-Welch . . . . 2. HMM. EM Baum- Welch 2 2. 1 HMM. Steve Young . 3 4 . Steve Young Baum-Welch . 4 . . . 3 4 1. . . 1 a, b, c . . . E-mail Jin'ichi Murakami, Member (Department of Information and Elec- . tronics, Graduate School of Engineering Tottori University 4-101, Koyamachou Minami, Tottori, 680-8552 Japan). . . Fundamentals Review 9 xxxx xx . c xxxx IEICE Fundamentals Review 1. 2. 3 HMM . Left-to-Right HMM. HMM 4 . . i i i . Left-to-Right HMM . 1 i = 0 0 = . HMM . HMM Left-to- HMM Right HMM . ai,j ai,j i j . . bi,j (O). bi,j (O) i j 2. 2 Ergodic HMM Left-to-Right HMM. O . parameter tie bi (O) . HMM Left-to-Right HMM Ergodic parameter tie 2.

図1 有限状態オートマトンの例 hmm は，確率付き有限状態オートマトンと同じパラメータ を持つ．しかし，確率付き有限状態オートマトンは，任意の時 刻に，どこの状態にいるのか明確に定義できるのに対し，hmm は，任意の時刻に，どの状態にいるのか定義できない．つまり，