Stochastic Gradient MCMC Methods for Hidden Markov Models
[edit]
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:22652274, 2017.
Abstract
Stochastic gradient MCMC (SGMCMC) algorithms have proven useful in scaling Bayesian inference to large datasets under an assumption of i.i.d data. We instead develop an SGMCMC algorithm to learn the parameters of hidden Markov models (HMMs) for timedependent data. There are two challenges to applying SGMCMC in this setting: The latent discrete states, and needing to break dependencies when considering minibatches. We consider a marginal likelihood representation of the HMM and propose an algorithm that harnesses the inherent memory decay of the process. We demonstrate the effectiveness of our algorithm on synthetic experiments and an ion channel recording data, with runtimes significantly outperforming batch MCMC.
Related Material


