Optimization-Based Control for the Extended Baum-Welch Algorithm
نویسندگان
چکیده
The extended Baum-Welch (EBW) is the most popular algorithm for discriminative training of speech recognition acoustic models. The EBW algorithm is usually controlled with heuristic rules, which are used to determine the smoothing parameters of the algorithm. In this paper we propose a control method for EBW which is based on the optimization of an error measure over a small control set. The large vocabulary speech recognition experiments show this to have clear benefits over the heuristic control.
منابع مشابه
Comparing the Bidirectional Baum-Welch Algorithm and the Baum-Welch Algorithm on Regular Lattice
A profile hidden Markov model (PHMM) is widely used in assigning protein sequences to protein families. In this model, the hidden states only depend on the previous hidden state and observations are independent given hidden states. In other words, in the PHMM, only the information of the left side of a hidden state is considered. However, it makes sense that considering the information of the b...
متن کاملGeneralized Baum-Welch and Viterbi Algorithms Based on the Direct Dependency among Observations
The parameters of a Hidden Markov Model (HMM) are transition and emission probabilities‎. ‎Both can be estimated using the Baum-Welch algorithm‎. ‎The process of discovering the sequence of hidden states‎, ‎given the sequence of observations‎, ‎is performed by the Viterbi algorithm‎. ‎In both Baum-Welch and Viterbi algorithms‎, ‎it is assumed that...
متن کاملGeneralization of extended baum-welch parameter estimation for discriminative training and decoding
We demonstrate the generalizability of the Extended BaumWelch (EBW) algorithm not only for HMM parameter estimation but for decoding as well. We show that there can exist a general function associated with the objective function under EBW that reduces to the well-known auxiliary function used in the Baum-Welch algorithm for maximum likelihood estimates. We generalize representation for the upda...
متن کاملComparison of optimization methods for discriminative training criteria
In this work we compare two parameter optimization techniques for discriminative training using the MMI criterion: the extended Baum-Welch (EBW) algorithm and the generalized probabilistic descent (GPD) method. Using Gaussian emission densities we found special expressions for the step sizes in GPD, leading to reestimation formula very similar to those derived for the EBW algorithm. Results wer...
متن کاملExtended Baum Transformations for General Functions, II
The discrimination technique for estimating the parameters of Gaussian mixtures that is based on the Extended Baum transformations (EB) has had significant impact on the speech recognition community. The proof that definitively shows that these transformations increase the value of an objective function with iteration (i.e., so-called ”growth transformations”) was presented by the author two ye...
متن کامل