Algebraic tests of general Gaussian latent tree models
Authors: Dennis Leung, Mathias Drton
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Numerical experiments in Section 4 makes comparisons to the likelihood ratio test and assesses the size of our tests in finite samples. We now report on some experiments with the bootstrap test based on the sup-norm of the estimated tetrads T proposed in Section 3. |
| Researcher Affiliation | Academia | Dennis Leung Department of Data Sciences and Operations University of Southern California dmhleung@uw.edu Mathias Drton Department of Statistics, University of Washington & Department of Mathematical Sciences, University of Copenhagen md5@uw.edu |
| Pseudocode | Yes | In conclusion, we perform the following multiplier bootstrap procedure: (i) Generate many, say E = 1000, sets of {e1, . . . , eω}, (ii) evaluate (3.8) for each of these E sets, and (iii) take q1 α to be the 1 α quantile from the resulting E numbers. |
| Open Source Code | No | The paper does not provide any specific links to source code, nor does it explicitly state that code is available in supplementary materials or elsewhere. |
| Open Datasets | No | Data are generated based on MX(T ) with parameters as prescribed in the text. We first consider two experimental setups, each with data generated from the one-factor model in (3.1) |
| Dataset Splits | No | The data used in the experiments is synthetically generated from a model (i.i.d. draws from a distribution), not from a larger dataset that would require explicit splits for training, validation, or testing. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used to run the experiments. |
| Software Dependencies | No | The paper mentions benchmarking against 'the function factanal in the base library of R' but does not specify the version number for R or the 'factanal' package. |
| Experiment Setup | Yes | In the implementation we always use E = 1000 sets of normal multipliers to simulate the quantile q1 α and work with batch size B = 3 in (3.8). ... The model parameters are as follows: (i) Setup 1: all loadings βp and error variances σ2 p,ϵ are taken to be 1. (ii) Setup 2: β1 and β2 are taken to be 10, while the other loadings are independently generated based on a normal distribution with mean 0 and variance 0.2. The error variances σ2 p,ϵ all equal 1/3. |