RISE: Robust Individualized Decision Learning with Sensitive Variables
Authors: Xiaoqing Tan, Zhengling Qi, Christopher Seymour, Lu Tang
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The reliable performance of the proposed method is demonstrated through synthetic experiments and three real-world applications. |
| Researcher Affiliation | Academia | Xiaoqing Tan University of Pittsburgh xit31@pitt.eduZhengling Qi George Washington University qizhengling@email.gwu.eduChristopher W. Seymour University of Pittsburgh seymourc@pitt.eduLu Tang University of Pittsburgh lutang@pitt.edu |
| Pseudocode | Yes | Algorithm 1: RISE (Robust individualized decision learning with sensitive variables) |
| Open Source Code | Yes | 1Python code is available at https://github.com/ellenxtan/rise. |
| Open Datasets | Yes | To illustrate the implication of the proposed method from a fairness perspective, we consider the National Supported Work (NSW) program [30] for improving personalized recommendations of a job training program on increasing incomes. ... To illustrate the implication of the proposed method from a safety perspective when there is delayed information, we consider the ACTG175 dataset among HIV positive patients [19]. |
| Dataset Splits | Yes | For simulation, we consider training data and testing data with sample sizes of 8,000 and 2,000, respectively. For real-data applications, we consider a 80-20 split of the dataset into a training data and a testing data. ... The details on modeling and hyperparameter tuning via cross-validations are given in Appendix C. |
| Hardware Specification | Yes | Experiments are performed on a 6-core Intel Xeon CPU E5-2620 v3 2.40GHz equipped with 64GB RAM. |
| Software Dependencies | No | The paper mentions software like 'neural networks' and 'Python package rise' and 'TensorFlow' but does not specify exact version numbers for these components. |
| Experiment Setup | Yes | The details on modeling and hyperparameter tuning via cross-validations are given in Appendix C. |