Pseudo-Extended Markov chain Monte Carlo

Authors: Christopher Nemeth, Fredrik Lindsten, Maurizio Filippone, James Hensman

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We compare the pseudo-extended method on three test models. The first two (Sections 4.1 and 4.2) are chosen to show how the pseudo-extended method performs on simulated data when the target is multi-modal. The third example (Section 4.3) is a sparsity-inducing logistic regression model, where multi-modality occurs in the posterior from three real-world datasets. We compare against popular competing algorithms from the literature, including methods discussed in Section 3.
Researcher Affiliation Collaboration Christopher Nemeth Department of Mathematics and Statistics Lancaster University United Kingdom c.nemeth@lancaster.ac.uk Fredrik Lindsten Department of Computer and Information Science Linköping University Sweden fredrik.lindsten@liu.se Maurizio Filippone Department of Data Science EURECOM France maurizio.filippone@eurecom.fr James Hensman PROWLER.io Cambridge United Kingdom james@prowler.io
Pseudocode Yes Several numerical integrators are available which preserve the volume and reversibility of the Hamiltonian system (Girolami and Calderhead, 2011), the most popular being the leapfrog integrator which takes L steps, each of size ϵ, though the Hamiltonian dynamics (pseudo-code is given in the Supplementary Material).
Open Source Code Yes 1https://github.com/chris-nemeth/pseudo-extended-mcmc-code
Open Datasets Yes We apply this model to three real-world data sets using micro-array data for cancer classification (prostate data results are given in Section E of the Supplementary Material, see Piironen and Vehtari (2017) for further details regarding the data).
Dataset Splits No The paper mentions 'held-out test data (random 20% of full data)' but does not explicitly provide training/validation/test dataset splits, specifically lacking details on a validation set or its proportion.
Hardware Specification No The paper does not provide specific hardware details such as exact GPU/CPU models, processor types, or memory amounts used for running its experiments.
Software Dependencies No In this paper, we use the No-U-turn sampler (NUTS) introduced by Hoffman and Gelman (2014) as implemented in the STAN (Carpenter et al., 2017) software package to automatically tune L and ϵ.
Experiment Setup Yes We set db = 28 (d = 27) and let (λ1, λ2) = (6, 2), as these settings have been shown to produce highly multi-modal distributions... Each sampler was run for 50,000 iterations (after burn-in) and the specific tuning details for the temperature ladder of PT and the energy rings for EE are given in Kou et al. (2006). In order to ensure a fair comparison between HMC and pseudo-extended HMC, we run HMC for 10,000 iterations and reduce the number of iterations of the pseudo-extended algorithms (with N = 2 and N = 5) to give equal total computational cost.