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].
Generalized Direct Change Estimation in Ising Model Structure
Authors: Farideh Fazayeli, Arindam Banerjee
ICML 2016 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results illustrating the effectiveness of the proposed estimator are presented. In this Section, we evaluate generalized direct change estimator (direct) with three different norms. and we compare our direct approach with indirect approach. For indirect approach, we first estimate Ising model structures ˆθ1 and ˆθ2 with L1 norm regularizer, separately (Ravikumar et al., 2010). Then, we obtain δˆθ = ˆθ1 ˆθ2. In all experiments, we draw n1 and n2 i.i.d samples from each Ising model by running Gibbs sampling. Here we set n = n1 = n2 = {20, 50, 100}. |
| Researcher Affiliation | Academia | Farideh Fazayeli EMAIL Arindam Banerjee EMAIL Department of Computer Science & Engineering, University of Minnesota, Twin Cities |
| Pseudocode | Yes | Algorithm 1 Generalized Direct Change Estimator |
| Open Source Code | No | The paper does not contain any explicit statement about releasing source code for the described methodology or a link to a code repository. |
| Open Datasets | No | In all experiments, we draw n1 and n2 i.i.d samples from each Ising model by running Gibbs sampling. Here we set n = n1 = n2 = {20, 50, 100}. This implies data was generated, but no access information is provided for a public dataset. |
| Dataset Splits | No | In all experiments, we draw n1 and n2 i.i.d samples from each Ising model by running Gibbs sampling. Here we set n = n1 = n2 = {20, 50, 100}. This describes sample generation for the models but does not specify a validation split. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used for the experiments. |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers (e.g., Python, PyTorch, or specific solver versions). |
| Experiment Setup | Yes | In all experiments, we draw n1 and n2 i.i.d samples from each Ising model by running Gibbs sampling. Here we set n = n1 = n2 = {20, 50, 100}. L1 norm: Here we first generate θ 1 with three disconnected star sub-graphs (Figure 4-a) with p = 50. We generate the weights uniformly random between {0.3 0.5}. We then generate θ 2 by removing 10 random edges from θ 1 (Figure 4-b). |