【正文】
A, B} ? ∏ = {pi} are the initial state probabilities ? A = {aij} are the state transition probabilities ? B = {bik} are the observation state probabilities A B A A A B B S S S K K K S K S K 36 HMM的應(yīng)用 (1) 評(píng)估 根據(jù)已知的 HMM找出一個(gè)觀察序列的概率 (2) 解碼 根據(jù)觀察序列找到最有可能出現(xiàn)的隱狀態(tài)序列 (3) 學(xué)習(xí) 從觀察序列中得出 HMM 37 HMM應(yīng)用 (1) ? 給定觀察序列 O=O1,O2,… OT,以及模型 , 計(jì)算 P(O|λ) ( , , )AB???oT o1 ot ot1 ot+1 38 HMM應(yīng)用 (1) ( | ) ( | , ) ( | )XP O P O X P X? ? ?? ?oT o1 ot ot1 ot+1 x1 xt+1 xT xt xt1 1 1 2 2( | , ) ... TTx o x o x oP O X b b b? ?1 1 2 2 3 1( | ) ... TTx x x x x x xP X a a a?? ??1 1 1 1 1 1111{ . . . }( | )t t t tTTx x o x x x otxxP O b a b??? ? ???? ? ?39 Forward Procedure ? 定義前向變量 ? 初始化: ? 遞歸: ? 終結(jié): 1( 1 ) 1i i i ob t T?? ? ? ? 11( 1 ) [ ( ) ] 1 1 , 1tNj i ij joit t a b t T j N????? ? ? ? ? ? ?? 1( / ) ( )NiiP O T???? ?12( ) ( , , , / ) 1i t tt P O O O x i t T??? ? ? ? 40 Forward Procedure )|,. ..()( 1 ?? ixooPt tti ??oT o1 ot ot1 ot+1 x1 xt+1 xT xt xt1 ?41 Forward Procedure 1)1( jojj b?? ?)1( ?tj?)|(),. . .()()|()|. . .()()|. . .(),. . .(1111111111111111jxoPjxooPjxPjxoPjxooPjxPjxooPjxooPtttttttttttttt?????????????????????????????????????????????????????????NijoijittttNitttttNitttttNittttbatjxoPixjxPixooPjxoPixPixjxooPjxoPjxixooP...1111...1111...11111...1111)()|()|(),. . .()|()()|,. . .()|(),. . .(?42 Backward Procedure ? 定義后向變量 ? 初始化: ? 遞歸: ? 終結(jié) : 12( ) ( , , , / ) 1 1t t t