Average Individual Fairness: Algorithms, Generalization and Experiments

Authors: Saeed Sharifi-Malvajerdi, Michael Kearns, Aaron Roth

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally we implement our algorithm and empirically verify its effectiveness. We have implemented the AIF-Learn algorithm and conclude with a brief experimental demonstration of its practical efficacy using the Communities and Crime dataset.
Researcher Affiliation Academia Michael Kearns University of Pennsylvania mkearns@cis.upenn.edu Aaron Roth University of Pennsylvania aaroth@cis.upenn.edu Saeed Sharifi-Malvajerdi University of Pennsylvania saeedsh@wharton.upenn.edu
Pseudocode Yes Subroutine 1: BEST best response of the Learner in the AIF setting. Algorithm 2: AIF-Learn learning subject to AIF. Mapping 3: b ψ (X, c W) pseudocode.
Open Source Code No The paper does not provide explicit statements about releasing source code for the described methodology or links to code repositories.
Open Datasets Yes We have implemented the AIF-Learn algorithm and conclude with a brief experimental demonstration of its practical efficacy using the Communities and Crime dataset. Described in detail and available for download at http://archive.ics.uci.edu/ml/datasets/ communities+and+crime
Dataset Splits No The paper mentions training and test sets but does not specify details for a separate validation dataset split.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments (e.g., CPU/GPU models, memory, or cloud instances).
Software Dependencies No The paper mentions using a "linear threshold learning heuristic" but does not specify any software names with version numbers for reproducibility.
Experiment Setup Yes To obtain a challenging instance of our multi-problem framework, we treated each of the first n = 200 neighborhoods as the individuals in our sample, and binarized versions of the first m = 50 variables as distinct prediction problems. Another d = 20 of the variables were used as features for learning. For the base learning oracle assumed by AIF-Learn, we used a linear threshold learning heuristic that has worked well in other oracle-efficient reductions (Kearns et al. (2018)).