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.
This issue may require collaboration/inspiration from https://github.com/Trusted-AI/AIF360/issues/341 Sometimes data may reside in cloud object store or amazon s3. Some companies may prefer to store "internal/confidential" data in a COS bucket....
MLFLow is an open source tool https://github.com/mlflow/mlflow/ that can help with various ML related tasks. The goal behind including MLFLow to AIf360 is that, as and when someone runs a...
Similar to issues: https://github.com/Trusted-AI/AIF360/issues/341 https://github.com/Trusted-AI/AIF360/issues/342 Goal is to have a simple way to download data from github - there are a few repo's that have "processed data" and we may...
Implement algorithm 2 in http://proceedings.mlr.press/v115/jiang20a/jiang20a.pdf - a simple and efficient post-processing fairness method
Create class/function similar to [RocCurveDisplay](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.RocCurveDisplay.html) which generates diagrams like in [Fig. 3 (Pleiss, et al. 2017)](https://arxiv.org/pdf/1709.02012.pdf). See the [example notebook](https://github.com/Trusted-AI/AIF360/blob/master/examples/sklearn/demo_new_features.ipynb) for a starting point.
- [ ] ~~Pass through `sample_weights` in `fit()`~~ - [ ] A few optional args are missing from our initializer (e.g. `grid_offset`) - [ ] Optional `random_state` arg for `__init__()`...
Add the [ThresholdOptimizer postprocessing algorithm](https://fairlearn.org/v0.5.0/api_reference/fairlearn.postprocessing.html) from Fairlearn into the sklearn-compat version of AIF360.
See eqn 10.5 in http://trustworthymachinelearning.com/trustworthymachinelearning-10.htm First priority is to do this for sklearn compatible version and then for "classic" version.
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.