Learning Fast-Mixing Models for Structured Prediction
This work addresses slow mixing issues in MCMC for structured prediction, which is a problem for researchers and practitioners in machine learning, but it appears incremental as it builds on existing MCMC methods with a specific parameterization.
The paper tackled the problem of slow mixing in Markov Chain Monte Carlo (MCMC) algorithms for structured prediction by introducing a new model family based on strong Doeblin Markov chains, which allows precise control over mixing times, and developed a learning algorithm that maximizes data likelihood under the stationary distribution, showing empirical improvements on two challenging inference tasks.
Markov Chain Monte Carlo (MCMC) algorithms are often used for approximate inference inside learning, but their slow mixing can be difficult to diagnose and the approximations can seriously degrade learning. To alleviate these issues, we define a new model family using strong Doeblin Markov chains, whose mixing times can be precisely controlled by a parameter. We also develop an algorithm to learn such models, which involves maximizing the data likelihood under the induced stationary distribution of these chains. We show empirical improvements on two challenging inference tasks.