Binary Bouncy Particle Sampler
This work addresses sampling challenges in binary probabilistic models, but it is incremental as it extends an existing method to a new domain.
The authors tackled the problem of sampling from binary distributions by generalizing the Bouncy Particle Sampler to handle piecewise differentiable cases, applying it to binary distributions with an augmentation method. In a binary Markov Random Field example, they found that binary BPS samplers perform better for easy-to-mix distributions compared to binary Hamiltonian Monte Carlo, though no concrete numbers were provided.
The Bouncy Particle Sampler is a novel rejection-free non-reversible sampler for differentiable probability distributions over continuous variables. We generalize the algorithm to piecewise differentiable distributions and apply it to generic binary distributions using a piecewise differentiable augmentation. We illustrate the new algorithm in a binary Markov Random Field example, and compare it to binary Hamiltonian Monte Carlo. Our results suggest that binary BPS samplers are better for easy to mix distributions.