Causal Inference with Non-IID Data using Linear Graphical Models
Authors: Chi Zhang, Karthika Mohan, Judea Pearl
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we test the coverage and performance of our methods through simulations. 5 Experiments 5.1 Simulations 5.2 Case Study |
| Researcher Affiliation | Academia | Chi Zhang Department of CS UCLA California, USA, 90095 zccc@cs.ucla.edu Karthika Mohan Department of EECS Oregon State University Oregon, USA, 97331 mohank@oregonstate.edu Judea Pearl Department of CS UCLA California, USA, 90095 judea@cs.ucla.edu |
| Pseudocode | Yes | Algorithm 1 Select a bias-free subset B from an interaction network G and return the largest subset from t iterations |
| Open Source Code | No | The paper does not provide an explicit statement about releasing the source code for the described methodology, nor does it provide a link to a code repository. |
| Open Datasets | No | The paper uses synthetic data generated through simulations, stating, 'We randomly generate balanced interaction network...' and 'We simulate data such that for each variable, the exogenous error term follows a Gaussian distribution with mean 0 and standard deviation 1.' It does not use or provide concrete access information for a publicly available or open dataset. |
| Dataset Splits | No | The paper describes experiments based on simulated data and evaluates the method's performance but does not specify any training, validation, or test dataset splits or cross-validation setup. |
| Hardware Specification | No | The paper describes its simulation experiments but does not provide specific details about the hardware used, such as GPU/CPU models, memory, or cloud resources. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers (e.g., Python, PyTorch, or specific solvers with their versions) that would be needed to replicate the experiments. |
| Experiment Setup | Yes | Simulated Model We randomly generate balanced interaction network with n units (i.e., the sample size is n), with Ci Xi Yi and Xi Mi for all i = 1, . . . , n. For all ordered pairs of distinct units i, j, we randomly add deflecting bias structures in the form of Xi Ci Yj with probability d Rate. For all units i, we randomly add reflecting bias structures in the form of Xi Mk Yi with probability r Rate for a random k = i. Simulation 1: Xi Yi s edge coefficient is 100, the edge coefficients of Ci Xi, Xi Mj, Mj Yi are all set to 10, the numbers of deflecting bias structures and additional reflecting bias structures are both 100. Simulation 2: Number of units n = 1000, Xi Yi s edge coefficient is 100, the numbers of deflecting bias structures and additional reflecting bias structures are both 100. |