Discrete Probabilistic Inverse Optimal Transport

Authors: Wei-Ting Chiu, Pei Wang, Patrick Shafto

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical contributions include visualizations of cross-ratio equivalent effect on basic examples, simulations validating theoretical results and experiments on real world data.
Researcher Affiliation Academia 1Department of Mathematics and Computer Science, Rutgers University Newark, NJ 2School of Mathematics, Institute for Advanced Study (IAS), Princeton NJ.
Pseudocode Yes Algorithm 1 Metro MC
Open Source Code No No explicit statement or link providing concrete access to open-source code for the methodology described in this paper was found.
Open Datasets Yes Application on European Migration data. We apply PIOT to analyze European migration flow. The observed migration data T g between 9 European countries for the period 2002-2007 is shown in Fig. 6 top (Raymer et al., 2013). All coupling Ts in this section are reported in Appendix B.9.
Dataset Splits No No specific training/validation/test dataset splits were explicitly provided. The paper discusses 'burn-in steps' and 'samples with lags' for MCMC, which relates to the sampling process, not data partitioning for model evaluation.
Hardware Specification No The experiments are performed on computing clusters with nodes equipped with 40 Intel Xeon 2.10 GHz CPUs and 192 GB memory.
Software Dependencies Yes We compare posterior sampled by MHMC to a m-dimensional uniform symmetric Dirichlet distribution generated by the Sci Py package (Virtanen et al., 2020).
Experiment Setup Yes We choose α = 1 for the Dirichlet prior P0(K), σ0 = 0.5, γ = 3, and δ = 1.0. We run for 10,000 burn-in steps and take 10,000 samples with lags of 100. We use σ = 10−6 for the burn-in steps and σ = 10−7.