Learning Unfaithful $K$-separable Gaussian Graphical Models

Authors: De Wen Soh, Sekhar Tatikonda

JMLR 2019 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We propose an algorithm to test the faithfulness of a conditional independence relation of the form Xu Xv | XS, where Xu, Xv and XS are the random variables associated with the nodes u, v and the node set S. This algorithm uses other conditional independence relations of the form Xi Xj | XS, where i, j / S to determine this. The faithfulness test does not require any assumption on the population version of covariance matrix Σ and can be applied to any Gaussian graphical model. To the best of our knowledge, this is the first algorithm that uses local information of a matrix to test for the faithfulness of a conditional independence relation. We also provide sample complexity bounds for this algorithm. We propose a structure learning algorithm for weakly K-separable Gaussian graphical models. [...] We propose a precision matrix learning algorithm for strongly K-separable Gaussian graphical models. This algorithm not only learns the structure of the graph, it learns the entries of the precision matrix (edge weights) as well. [...] In this section, we will examine some examples where our algorithm can perform structural estimation while others that rely on sparsity cannot.
Researcher Affiliation Academia De Wen Soh EMAIL Institute of High Performance Computing 1 Fusionopolis Way, #16-16 Connexis Singapore, 138632 Sekhar Tatikonda EMAIL Department of Electrical Engineering Yale University New Haven, CT 06511, USA
Pseudocode Yes Algorithm 1: Testing Faithfulness of Relation Xu Xv | XS Algorithm 2: Learning topology of weakly K-separable GGM Algorithm 3: Learning the precision matrix of strongly K-separable GGM
Open Source Code No The paper does not provide any explicit statements about releasing code, nor does it provide links to a code repository or mention code in supplementary materials.
Open Datasets No The paper uses synthetic examples, such as '4-dimensional Gaussian distribution' and '6-dimensional Gaussian distribution', but does not mention or provide access information for any publicly available or open datasets.
Dataset Splits No The paper is theoretical and does not conduct experiments on datasets, therefore, it does not specify any dataset splits.
Hardware Specification No The paper is theoretical and focuses on algorithm design and theoretical guarantees. It does not describe any experiments that would require specific hardware, thus no hardware specifications are provided.
Software Dependencies No The paper is theoretical and focuses on algorithm design. It does not mention any specific software dependencies with version numbers that would be required to reproduce experiments.
Experiment Setup No The paper focuses on theoretical algorithms and their properties, rather than empirical experiments. Therefore, it does not provide details on experimental setup, hyperparameters, or training configurations.