Laplacian Hamiltonian Monte Carlo

Yizhe Zhang, Changyou Chen, Ricardo Henao, Lawrence Carin

Abstract

We proposed a Hamiltonian Monte Carlo (HMC) method with Laplace kinetic energy, and demonstrate the connection between slice sampling and proposed HMC method in one-dimensional cases. Based on this connection, one can perform slice sampling using a numerical integrator in an HMC fashion. We provide theoretical analysis on the performance of such sampler in several univariate cases. Furthermore, the proposed approach extends the standard HMC by enabling sampling from discrete distributions. We compared our method with standard HMC on both synthetic and real data, and discuss its limitations and potential improvements.
BibTeX
@inproceedings{zhang2016laplacian,
  author = {Yizhe Zhang and Changyou Chen and Ricardo Henao and Lawrence Carin},
  title = {{Laplacian Hamiltonian Monte Carlo}},
  booktitle = {ECML/PKDD},
  year = {2016},
  doi = {10.1007/978-3-319-46128-1\_7},
}

Loading related content...

Figure