AIF360
AIF360 copied to clipboard
Trusted-AI
A comprehensive set of fairness metrics for datasets and machine learning models, explanations for these metrics, and algorithms to mitigate bias in datasets and models.
According to Theil index definition, the formula would be undertemined when y_score is 0 and y_target is 1. Since b_i = y_score - y_target + 1 would give b_i =...
Line 114 of the attached .py file leads to a memory leak and crash after several thousands of iterations  when called from the attached notebook. The error is resolved...
Line 70 of the attached .py file leads to a memory leak and crash after several thousands of iterations  when called from the attached notebook. The error is resolved...
Users may want to try AIF360 inside a docker container to launch on Kubernetes/Openshift environments.
This PR solves #263. It provides users the ability to run AIF360 inside a docker container by adding datasets inside the container. This will be helpful to users launching containers...
The `PrejudiceRemover` inprocessing bias mitigation algorithm uses `LogisticRegression`, and by default, `max_iter` for `LogisticRegression` from sklearn is 100. However, in practice, more iterations are occasionally needed to fit `PrejudiceRemover` to...
I'm trying to use the adversarial debiasing method but always getting this issue: ``` Variable debiased_classifier/classifier_mode/W1 already exists, disallowed. Did you mean to set reuse-True... ``` I suggest using something...
In AIF360, age>=25 is set as the privileged class in BankDataset. However, I find that this class is not really privileged. In fact, 12.35% of samples with age>=25 were marked...
This issue is to track regression metric addition to the toolkit. - [ ] Mean Absolute Error
Inequality indices measure how unevenly the benefits of an algorithm are distributed. In the case of a binary classifier, all algorithmic predictions (where the ground truth is known) can be...
Metadata
Owner
Metadata
A comprehensive set of fairness metrics for datasets and machine learning models, explanations for these metrics, and algorithms to mitigate bias in datasets and models.