On the Impossibility of Non-trivial Accuracy in Presence of Fairness Constraints
Authors: Carlos Pinzón, Catuscia Palamidessi, Pablo Piantanida, Frank Valencia7993-8000
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Theoretical | The paper focuses on theoretical proofs, propositions, and the characterization of properties of data sources and predictors (e.g., Theorem 1, Theorem 3, Proposition 6, Proposition 7). It includes illustrative figures generated from theoretical models (e.g., Figure 1, Figure 2, Figure 4) and provides algorithms for generating such examples, but does not present empirical studies with real-world datasets, experimental evaluations, or performance metrics from trained models. |
| Researcher Affiliation | Academia | 1Inria, Paris-Saclay, France 2CNRS, France 3Laboratoire d informatique de l École Polytechnique (LIX) 4Laboratoire des Signaux et des Systèmes (L2S), Université Paris-Saclay, Centrale Supélec 5Pontificia Universidad Javeriana Cali, Colombia |
| Pseudocode | Yes | Algorithm 1: Random generator for Theorem 3. |
| Open Source Code | Yes | The Python code for generating this figure is in the supplementary material. Algorithm 1 (Python code in supplementary material). |
| Open Datasets | No | The paper is theoretical and discusses properties of data sources and predictors using illustrative examples (e.g., 'Example 1. Consider a data source (X, A, Y ) over {0, 1}3 whose distribution is given by...'). It does not use external datasets for training machine learning models, and therefore does not provide access information for a training dataset. |
| Dataset Splits | No | The paper is theoretical and does not involve empirical training, validation, or testing splits on real datasets. It does not provide any information about dataset splits. |
| Hardware Specification | No | The paper is theoretical and does not involve empirical experiments requiring specific hardware. Therefore, no hardware specifications are mentioned. |
| Software Dependencies | No | The paper mentions 'Python code' for generating figures and algorithms, but it does not specify any version numbers for Python or any specific libraries or software dependencies required for reproducibility. |
| Experiment Setup | No | The paper is theoretical and does not describe empirical experiments with hyperparameters, model initialization, or training schedules. Therefore, no experimental setup details are provided. |