Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Differentially Private Empirical Risk Minimization under the Fairness Lens

Authors: Cuong Tran, My Dinh, Ferdinando Fioretto

NeurIPS 2021 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The proposed approach is evaluated on several datasets and settings.
Researcher Affiliation Academia Cuong Tran Syracuse University EMAIL My H. Dinh Syracuse University EMAIL Ferdinando Fioretto Syracuse University EMAIL
Pseudocode Yes Algorithm 1: DP-SGD input :Disjoint dataset D ; Sample prob. q; Iterations T; Noise variance σ2; Clipping bound C; learning rate η
Open Source Code No The paper does not provide an explicit statement about the release of source code for the methodology described, nor does it include a direct link to a code repository.
Open Datasets Yes The proposed approach is evaluated on several datasets and settings. ... on two datasets. Each data point represents the average of 100 runs of a DP Logistic Regression (obtained with output perturbation) on each group z A. Details on dataset and experimental setting are provided in Appendix B and additional experiments in Appendix C. ... Healthcare dataset stroke data. URL http://www.kaggle.com/fedesoriano/ stroke-prediction-dataset. ... UCI repository of machine learning databases, 1988. URL https: //archive.ics.uci.edu/ml/datasets.php. ... Telco customer churn dataset, 2015. URL http://www.ibm.com/communities/analytics/watson-analyticsblog/ predictive-insights-in-the-telco-customer-churn-data-set/.
Dataset Splits No The paper mentions 'Details on dataset and experimental setting are provided in Appendix B and additional experiments in Appendix C.' but does not explicitly provide specific training/test/validation dataset splits in the main text.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4).
Experiment Setup Yes Algorithm 1: DP-SGD input :Disjoint dataset D ; Sample prob. q; Iterations T; Noise variance σ2; Clipping bound C; learning rate η ... The experiment use C = 0.1 and σ = 1. ... The implementation uses a neural network with a single hidden layer and Suppose uses DP-SGD with C = 0.1, σ = 5.0.