Gibbsian Polar Slice Sampling

Authors: Philip Schär, Michael Habeck, Daniel Rudolf

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments in a variety of settings indicate that our proposed algorithm outperforms the two most closely related approaches, elliptical slice sampling (Murray et al., 2010) and hit-and-run uniform slice sampling (Mac Kay, 2003).
Researcher Affiliation Academia 1Microscopic Image Analysis Group, Friedrich Schiller University Jena, Jena, Germany 2Faculty of Computer Science and Mathematics, University of Passau, Passau, Germany.
Pseudocode Yes Algorithm 1 Gibbsian Polar Slice Sampling; Algorithm 2 Geodesic Shrinkage; Algorithm 3 Radius Shrinkage
Open Source Code Yes Source code allowing the reproduction (in nature) of our experimental results is provided in a github repository6. (Footnote 6: https://github.com/microscopic-imageanalysis/Gibbsian Polar Slice Sampling)
Open Datasets Yes We consider the Cover Type data set (Blackard, 1998; Blackard et al.)
Dataset Splits Yes We use only 10% of it as training data and the remaining 90% as test data.
Hardware Specification Yes All of our experiments were conducted on a workstation equipped with an AMD Ryzen 5 PRO 4650G CPU.
Software Dependencies Yes an easily usable, general purpose implementation of GPSS in Python 3.10, based on numpy.
Experiment Setup Yes We chose the sample space dimension to be d = 100, initialized all samplers with x0 := (1, . . . , 1)T and ran each of them for N = 106 iterations.