Forward-backward algorithm

The forward-backward algorithm can be used to find the most likely state for any point in time. It cannot, however, be used to find the most likely sequence of states (see Viterbi algorithm).
[edit] Forward Probabilities ...

• Statistical sequence labeling - Features a comprehensive introduction to hidden Markov models (HMMs) including the forward, Viterbi, and forward-backward algorithms, maximum entropy models, and maximum entropy Markov models (MEMMs).

To be labelled on a phone-by-phone basis rather than a sentence-by-sentence basis, and that too is a difficult task. The standard algorithm for training an HMM is called the Baum-Welch or forward-backward algorithm.

See also: Speech recognition, Smoothing, Inference, Knowledge, Artificial intelligence