The Multiple Quantile Graphical Model

Authors: Alnur Ali, J. Zico Kolter, Ryan J. Tibshirani

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Lastly, we present detailed experiments that demonstrate the flexibility and effectiveness of the MQGM in modeling hetereoskedastic non-Gaussian data. (from Abstract) and Section 6 presents empirical examples comparing the MQGM versus common alternatives. (from Section 1 Introduction)
Researcher Affiliation Academia Alnur Ali Machine Learning Department Carnegie Mellon University alnurali@cmu.edu J. Zico Kolter Computer Science Department Carnegie Mellon University zkolter@cs.cmu.edu Ryan J. Tibshirani Department of Statistics Carnegie Mellon University ryantibs@cmu.edu
Pseudocode No The paper describes the computational approach using ADMM in Section 5, providing equations and a description of updates, but it does not include a structured pseudocode block or algorithm box.
Open Source Code No The paper does not include an explicit statement about releasing source code or provide a link to a code repository for the methodology described.
Open Datasets Yes We study n = 937 weekly flu incidence reports from September 28, 1997 through August 30, 2015, across 10 regions in the United States (see the top panel of Figure 2), obtained from [6]. [6] Centers for Disease Control and Prevention (CDC). Influenza national and regional level graphs and data, August 2015. URL http: //gis.cdc.gov/grasp/fluview/fluportaldashboard.html.
Dataset Splits No The paper mentions sample sizes for synthetic and flu data (e.g., 'n = 400 samples', 'n = 937 weekly flu incidence reports') but does not provide specific details on how these samples were split into training, validation, and test sets for reproducibility.
Hardware Specification Yes All subproblems took about 1 minute on a 6 core 3.3 Ghz Core i7 X980 processor.
Software Dependencies No The paper mentions software like ADMM and SCS for solving the problem but does not provide specific version numbers for these or other key software components used in the experiments.
Experiment Setup Yes First fix some ordered set A = {α1, . . . , αr} of quantile levels, e.g., A = {0.05, 0.10, . . . , 0.95}. For each variable yk, and each level αℓ, we model the conditional αℓ-quantile... Here λ1, λ2 0 are tuning parameters, Fℓkj, j = 1, . . . , d are univariate function spaces, ω > 0 is a fixed exponent... The MQGM was used with m = 10 basis functions (RBFs), and r = 20 quantile levels... We set m = 5, r = 99, and also introduced exogenous variables to encode the week numbers, so p = 1.