Table 1 The computational expense of different algorithm implementations running on HMM.
From: Implementing EM and Viterbi algorithms for Hidden Markov Model in linear memory
Algorith | Canonical | Checkpointing | Linear | |||
|---|---|---|---|---|---|---|
Viterbi | Time | O(TNQ max ) | Time | O(TNQ max ) | Time | O(TNQ max ) |
Space | O(TN) | Space |
| Space | O(T) | |
Baum-Welch | Time | O(TNQ max + T (Q + E)) | Time | O(TNQ max + T (Q + E)) | Time | O(TNQ max (Q + ED)) |
Space | O(TN) | Space |
| Space | O(N(Q + ED)) | |