Viterbi Algorithm
A dynamic-programming algorithm that finds the most probable hidden-state sequence in a model with Markov structure and local transition scores.
Search-Algorithm Context
The Viterbi algorithm uses dynamic programming to keep the best predecessor score for each state at each time step, then follows stored backpointers to recover the highest-scoring complete path. This avoids enumerating every possible hidden-state sequence.
Inference Boundary
Viterbi returns one best path. If the goal is posterior marginal probability over states or transitions, forward-backward inference serves a different purpose.
Related Search Algorithms
- Hidden Markov Model
- Dynamic Programming
- Beam Search
- Trellis