Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
On Semiparametric Exponential Family Graphical Models
Authors: Zhuoran Yang, Yang Ning, Han Liu
JMLR 2018 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section we study the finite-sample performance of the proposed graph inference methods on both simulated and real-world datasets. 5.1. Simulation Studies We first examine the numerical performance of the proposed pairwise score tests for the null hypothesis H0 : β jk = 0. We simulate data from the following three settings: (i) Gaussian graphical model. (ii) Ising graphical model. (iii) Mixed graphical model. 5.2. Real Data Analysis We then apply the proposed methods to analyze a publicly available dataset named Computer Audition Lab 500-Song (CAL500) dataset (Turnbull et al., 2008). |
| Researcher Affiliation | Academia | Zhuoran Yang EMAIL Department of Operations Research and Financial Engineering Princeton University Princeton, NJ 08544, USA Yang Ning EMAIL Department of Statistical Science Cornell University Ithaca, NY 14853, USA Han Liu EMAIL Department of Electrical Engineering and Computer Science Northwestern University Evanston, IL 60208, USA |
| Pseudocode | Yes | We present the proposed adaptive multi-stage convex relaxation method in Algorithm 1. Algorithm 1 Adaptive Multi-stage Convex Relaxation algorithm for parameter estimation 1: Initialize λ(0) jk = λ for 1 j, k d. 2: for j= 1,2,. . . ,d do 3: for ℓ= 1, 2, . . . , until convergence do 4: Solve the convex optimization problem bβ(ℓ) j = argmin Rd 1 n Lj(βj) + X k =j pλ(|βjk|) o . |
| Open Source Code | No | The paper mentions using the PICASSO package (Ge et al., 2017) and the Ising Sampler package (Epskamp, 2015), which are third-party tools. However, there is no explicit statement or link indicating that the authors have open-sourced their own implementation code for the methodology described in this paper. |
| Open Datasets | Yes | We then apply the proposed methods to analyze a publicly available dataset named Computer Audition Lab 500-Song (CAL500) dataset (Turnbull et al., 2008). The data can be obtained from the Mulan database (Tsoumakas et al., 2011). |
| Dataset Splits | No | The paper mentions `n = 100` and `d = 200` for simulation studies and `n = 502` and `d = 226` for real data analysis, specifying the overall dataset sizes. It states, "The parameter λ is chosen by 10-fold cross validation as suggested by Ning et al. (2017b)." While 10-fold cross-validation is mentioned for parameter selection (for `λ`), there are no explicit details about how the datasets themselves are split into training, validation, or test sets for evaluating model performance. |
| Hardware Specification | No | The paper does not explicitly describe any specific hardware used to run the experiments, such as CPU or GPU models, memory specifications, or types of computing clusters. |
| Software Dependencies | No | The paper mentions the use of 'the PICASSO package (Ge et al., 2017)' and 'Ising Sampler (Epskamp, 2015)' for certain tasks, but it does not provide specific version numbers for these or any other software dependencies. There is also a mention of 'R package' for Ising Sampler, but again, no version. |
| Experiment Setup | Yes | We simulate data from the following three settings: (i) Gaussian graphical model. We set n = 100 and d = 200. The graph structure is a 4-nearest-neighbor graph... We set Θjj = 1, Θjk = µ [0, 0.25)... (ii) Ising graphical model. We set n = 100 and d = 200... We set |β jk| = µ [0, 1]... (iii) Mixed graphical model. We set n = 100 and d = 200... We set |β jk| = µ [0, 1]... The parameter λ is chosen by 10-fold cross validation... We set the nonconvex penalty function in optimization problem (7) to be capped-ℓ1 penalty pλ(u) = λ min{u, λ} with the regularization parameter λ selected by 10-fold cross-validation... |