Relaxations for inference in restricted Boltzmann machines
This work addresses inference challenges in machine learning models like restricted Boltzmann machines, but it appears incremental as it builds on existing relaxation and sampling techniques.
The paper tackles the problem of approximate inference in restricted Boltzmann machines by proposing a relaxation-based algorithm to sample near-MAP configurations and estimate the log-partition function, showing comparisons against other sampling-based methods.
We propose a relaxation-based approximate inference algorithm that samples near-MAP configurations of a binary pairwise Markov random field. We experiment on MAP inference tasks in several restricted Boltzmann machines. We also use our underlying sampler to estimate the log-partition function of restricted Boltzmann machines and compare against other sampling-based methods.