Tsallis Regularized Optimal Transport and Ecological Inference

Authors: Boris Muzellec, Richard Nock, Giorgio Patrini, Frank Nielsen

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate empirically the TROT framework with its application to ecological inference. The dataset we use describes about 10 millions individual voters from Florida for the 2012 US presidential elections, as obtained from (Imai and Khanna 2016). As a demonstrative example, we focus on inferring the distributions of ethnicity and party for all Florida counties. ... Table 2 reports a quantitative comparison.
Researcher Affiliation Collaboration Ecole Polytechnique, France; Data61, Australia; The Australian National University, Australia The University of Sydney, Australia; Sony CS Labs, Inc., Japan
Pseudocode Yes Algorithm 1 Second Order Row TROT (SO TROT) ... Algorithm 2 KL Projected Gradient TROT (KL TROT)
Open Source Code No No explicit statement or link providing open-source code for the described methodology.
Open Datasets Yes The dataset we use describes about 10 millions individual voters from Florida for the 2012 US presidential elections, as obtained from (Imai and Khanna 2016).
Dataset Splits Yes First, the ground truth joint distributions for one district are known; we chose district number 3 which groups 9 out of 68 counties of about 285K voters in total. This information will be used to tune hyper-parameters.
Hardware Specification No No specific hardware details for running experiments are provided.
Software Dependencies No No specific software dependencies with version numbers are provided.
Experiment Setup Yes We study the solution of TROT for a grid of λ [0.01, 1000], q [0.5, 4], inferring the joint distributions of all counties of district number 3. ... The dissimilarity measure relies on a Gaussian kernel between average county profiles: m RBF ij .= 2 2 exp( γ μp i μe j 2) , (11) with γ = 10.