Minimax Pareto Fairness: A Multi Objective Perspective

Authors: Natalia Martinez, Martin Bertran, Guillermo Sapiro

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test the proposed methodology on real case-studies of predicting income, ICU patient mortality, skin lesions classification, and assessing credit risk, demonstrating how our framework compares favorably to other approaches.
Researcher Affiliation Academia 1Department of Electircal and Computer Engineering, Duke University.
Pseudocode Yes Algorithm 1 APStar
Open Source Code Yes Code is available at github.com/natalialmg/MMPF.
Open Datasets Yes In addition to this, we demonstrate how our methodology performs on real tasks such as inferring income status in the Adult dataset (Dua & Graff, 2017a), predicting ICU mortality rates in the MIMIC-III dataset from hospital notes (Johnson et al., 2016), classifying skin lesions in the HAM10000 dataset (Tschandl et al., 2018), and assessing credit risk on the German Credit dataset (Dua & Graff, 2017b).
Dataset Splits No The paper mentions 'standard deviations computed across 5 splits' but does not provide specific percentages or details about a dedicated validation set used for hyperparameter tuning or model selection.
Hardware Specification No The paper mentions training DNNs and neural networks but does not provide specific details on the hardware used, such as GPU/CPU models, memory, or cloud instances.
Software Dependencies No The paper mentions 'Pytorch code' and 'Stochastic Gradient Descent' (SGD) but does not specify version numbers for these or other software dependencies.
Experiment Setup Yes Classifiers are implemented using neural networks and/or linear logistic regression; for details on architectures and hyperparameters, refer to Section A.9.