1. ngram-1) 13: if ? node �𝑙 ? 𝑛 𝑘 .expanded nodes such that �𝑙 .hist = hist then 14: �𝑙 ? 𝑎 𝑘 .Duplicate(� 𝑗 , �𝑙 ) ? Create arc between expanded nodes 15: else 16: �𝑙 ? 𝑛 𝑘 .Duplicate(hist) ? Create expanded node with new history 17: 𝑛 𝑘 .expanded nodes.Append(� 𝑙 ) 18: �𝑙 ? 𝑎 𝑘 .Duplicate(� 𝑗 , �𝑙 ) ? Create arc between expanded nodes 19: end if 20: post ? LastNItems([cache.GetPost(� 𝑗 .hist, � 𝑗 .time), �𝑙 .post], ngram-1) 21: if cache;K 𝑛 𝑖 ; Cho;Assign score from label-sync model 29: �𝑙 .model score ? cache.GetPred(� 𝑗 .hist, � 𝑗 .time, �𝑙 .word) 30: ? Connect new arc to expanded nodes 31: � 𝑗 .exits.Append(� 𝑙 ); �𝑙 .entries.Append(� 𝑙 ),2015
2. An in equality with applications to statistical estimation for probabilistic functions of a markov process and to a model for ecology;L Baum;Bull. of the Am. Meteorol. Soc,1967
3. Connectionist speech recognition: a hybrid approach;H A Bourlard;Springer Sci. Bus. Media,1994