A new algorithm for inference in Hidden Markov models with lower span complexity
15/05/2024 Wednesday 15th May 2024, 14:30 (Room P3.10, Mathematics Building)
More
Diogo Pereira, CEMAT, Instituto Superior Técnico
The maximum likelihood problem for Hidden Markov Models is usually numerically solved by the BaumWelch algorithm, which uses the ExpectationMaximization algorithm to find the estimates of the parameters. This algorithm has a recursion depth equal to the data sample size and cannot be computed in parallel, which limits the use of modern GPUs to speed up computation time. A new algorithm is proposed that provides the same estimates as the BaumWelch algorithm, requiring about the same number of iterations, but is designed in such a way that it can be parallelized. As a consequence, it leads to a significant reduction in the computation time. We illustrate this by means of numerical examples, where we consider simulated data as well as real datasets.
