Optimal Treatment Regimes for Proximal Causal Learning
Authors: Tao Shen, Yifan Cui
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Furthermore, we demonstrate the proposed optimal regime via numerical experiments and a real data application. |
| Researcher Affiliation | Academia | Tao Shen National University of Singapore Yifan Cui Zhejiang University |
| Pseudocode | No | The paper describes the estimation steps in text but does not provide any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any explicit statement about making its source code publicly available or a link to a code repository for the methodology described. |
| Open Datasets | Yes | The data generating mechanism for (X, A, Z, W, U) follows the setup proposed in Cui et al. (2023) and is summarized in Appendix I. ... We consider six scenarios in total, and the setups of varying parameters are deferred to Appendix I. For each scenario, training datasets {Yi, Ai, Xi, Zi, Wi}n i=1 are generated following the above mechanism with a sample size n = 1000. ... The data have been re-analyzed in a number of papers in both causal inference and survival analysis literature ... under the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT, Connors et al. (1996)). |
| Dataset Splits | Yes | For each scenario, training datasets {Yi, Ai, Xi, Zi, Wi}n i=1 are generated following the above mechanism with a sample size n = 1000. ... The testing dataset is generated with a size 10000, and the empirical value function for the estimated ITR is used as a performance measure. ... Gaussian kernel is used to estimate ˆπ, with bandwidth γ chosen from {0.1, 0.2, ..., 1.0} using cross-validation. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running the experiments (e.g., CPU/GPU models, memory). |
| Software Dependencies | No | The paper mentions software components like 'PyTorch' and the use of 'logistic regression with L2 regularization', 'Adam optimizer', and 'Gaussian kernel', but does not provide specific version numbers for these software dependencies. |
| Experiment Setup | Yes | In particular, the preliminary ITRs ˆdz and ˆdw are estimated using a linear decision rule, and ˆπ(x; ˆdz, ˆdw) is estimated using a Gaussian kernel. More details can be found in the Appendix K. ... The neural network consists of 5 fully-connected hidden layers with 100 neurons each, and leaky ReLU activation functions. We use the Adam optimizer with a learning rate of 1e-4 and a batch size of 256 for 1000 epochs. |