inFairness
inFairness copied to clipboard

Published 20 hours ago •

→

Metadata

PyTorch package to train and audit ML models for Individual Fairness

Reame
Issues

Results 19 inFairness issues

Sort by recently updated

check if we can use the code from a PR in pytorch

1

comment

lr_factor in the ranking algorithm should work stably, should explore their values before using the current heuristic.

expose API to sample from a trained ranking neural net, both maximum likelihood and plackett-luce sampling.

Also, the output prediction should include the rankings, not only the scores

add ranking auditor capability to use loss ratio

speed-up comparable data pair generation method

method link: https://github.com/IBM/inFairness/blob/main/inFairness/utils/datautils.py#L9

enhancement

add experiments on retiring adults dataset

retiring adults dataset: https://github.com/zykls/folktables

include group fairness metrics in the package

enhancement

help wanted

add individually fair ranking algo to the package

paper link: https://arxiv.org/abs/2103.11023

enhancement

support continuous and discrete protected attributes for Logistic Regression metric

enhancement

← Metadata

61

Stars

7

Forks

Watchers

Owner

Metadata

PyTorch package to train and audit ML models for Individual Fairness