Tier Balancing: Towards Dynamic Fairness over Underlying Causal Factors
Authors: Zeyu Tang, Yatong Chen, Yang Liu, Kun Zhang
ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we present experimental results on both simulated and real-world FICO data set from Board of Governors of the Federal Reserve System (US) (2007). In the sequence of decision-distribution interplay, the latent causal factor HT is updated according to the specified dynamics (Equation 3) at each time step. The output of the decision policy (at each time step) depends on the specific scenario. In particular, we consider perfect predictors and Counterfactual Fair (Kusner et al., 2017, Level 1 implementation) predictors. |
| Researcher Affiliation | Academia | 1Department of Philosophy, Carnegie Mellon University 2Computer Science and Engineering Department, University of California, Santa Cruz 3Machine Learning Department, Mohamed bin Zayed University of Artificial Intelligence |
| Pseudocode | No | The paper does not include pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | Yes | Our code repository is available on Github: https://github.com/zeyutang/Tier Balancing. |
| Open Datasets | Yes | real-world FICO data set from Board of Governors of the Federal Reserve System (US) (2007). |
| Dataset Splits | No | The paper mentions using the 'preprocessed credit score data set (Hardt et al., 2016)' and converting 'CDF... into group-wise density distributions... and use them as the initial tier distributions'. However, it does not explicitly provide specific training, validation, or test dataset splits (e.g., percentages or sample counts). |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU/CPU models, memory, or types of computing resources used for the experiments. |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers (e.g., programming languages, libraries, or frameworks with their versions). |
| Experiment Setup | No | The paper states, 'The output of the decision policy (at each time step) depends on the specific scenario. In particular, we consider perfect predictors and Counterfactual Fair (Kusner et al., 2017, Level 1 implementation) predictors.' and 'The Counterfactual Fair decision-making strategy is retrained after each one-step data dynamics.' While these describe aspects of the experimental setup, they do not provide concrete numerical hyperparameters (like learning rates, batch sizes, or optimizer settings) or a detailed, titled section explicitly for experiment setup as required. |