Content deleted Content added
Added an overview section and a semi formal description of the algorithm based on matrices as well as referenced a few known performance enhancements. Added a numerical example. |
|||
Line 114:
==Performance==
The brute force procedure for the solution of this problem is the generation of all possible sequences of observed events and hidden states with their probabilities using the two transition matrices. The joint probability of two sequences, given the model, is calculated by multiplying the corresponding probabilities. The brute force procedure has [[time complexity]] <math> O(T \cdot N^T) </math>, where <math>T</math> is the length of sequences and <math>N</math> is the number of symbols in the state alphabet. This is intractable for realistic problems, as the number of possible hidden node sequences typically is extremely high. However, the forward-backward algorithm has time complexity <math> O(N^2 T)\, </math>.
Several enhancements are known to the general forward-backward algorithm which allow for computations to take place in constant space. In addition, as t grows, algorithms have been developed to compute <math>\mathbf{f_{1:t+1}}</math> efficiently through online
==See also==
| |||