Frequently asked questions

Where can I learn more about fairness in machine learning?

Please review further resources, where we provide links to various materials that we have found helpful. Also, in our user guide, we link to the papers describing the algorithms implemented in Fairlearn.

Why not just ignore the sensitive features?

If your goal is to train a model whose predictions are statistically independent of the sensitive features, then it is not enough to simply ignore the sensitive features. Information is often redundantly encoded across several features, and machine learning algorithms will uncover these links (it is what they are designed to do). For example, in the US, the ZIP code where a person lives is well correlated with their race. Even if the model is not provided with race as a feature, the model will pick up on it implicitly via the ZIP code (and other features). Worse, without having the race available in the dataset, it is hard to assess the model’s impact across different groups defined by race or by race intersected with other demographic features.

The model is unfair because the data are biased. Isn’t it better to get better data?

The answer to this question depends on what is meant by ‘unfair’, ‘biased data’, and ‘better data’ in any particular context. Consider the example of a company seeking to build a tool for screening the resumes of job candidates. The company is planning to use their internal job evaluation data and train a model to predict job evaluations of the applicants; those with higher predictions will be ranked higher by the screening tool. This setup might present several fairness issues:

  • If the company has historically hired few women, there will be fewer of them in the training data set, and so a trained model may be less accurate for them.

  • The choice of features also affects the accuracy of the model. The features that are predictive for one group of applicants might not be as predictive for another group, and so more data will not necessarily improve the accuracy.

  • The accuracy of a model might not mean that the model is fair. If women have received systematically poorer reviews due to biased managers or worse workplace conditions, then the model might appear to be accurate, but the choice of the label (in this case, job evaluation) does not accurately reflect the applicants’ potential.

These are just three ways how the data may be ‘biased’, and they are not mutually exclusive. The processes for getting ‘better data’ will be different for each. In some of these cases, obtaining ‘better data’ may not be practical, but it might still be possible to use some mitigation algorithms.

Why am I seeing fairness issues, even though my data are reflective of the general population?

Machine learning models often perform poorly for subgroups which are poorly represented. What constitutes poor representation is context specific, and may well be affected by historical misrepresentation (consider the example above, of a company which had previously hired few women). For this reason, balanced sampling is generally better for ML than population sampling. On a related point, this is why it is important to consider multiple fairness metrics, and how they vary across different subgroups.

Won’t making a model fairer reduce its accuracy?

There are often many machine learning models that achieve similar levels of accuracy or other performance metrics, but that dramatically differ in how they affect different subgroups. Mitigation algorithms seek to improve the fairness metrics without strongly affecting the accuracy, or more generally to navigate the trade-offs between performance and fairness metrics.

Can the mitigation algorithms in Fairlearn make my model fair?

There are many ways in which a model can be unfair. Fairlearn mitigation algorithms only address some of them: those that can be quantified by our supported fairness metrics. However, to assess whether the new model is fairer, it is important to consider not only the fairness metrics, but also the societal and technical context in which the model is applied.

I’ve got improved data, trained and mitigated new models, checked all the metrics… am I done?

Firstly, always remember that there is more to fairness than technical details such as metrics - fairness is a sociotechnical problem. Even if these are all considered at training time, the ML lifecycle doesn’t end when a model is deployed. Models need to be monitored in production. On a technical level, this means checking for data drift within the vulnerable subgroups identified during the fairness analysis. However the societal aspects need to be considered as well, for example:

  • Are the actual harms as expected, both in the nature of the harms and their distribution?

  • Have the users of the model (who may not be the subjects of the model) adjusted their usage patterns? This is sometimes called ‘strategic behavior.’

What sort of fairness-related harms can the Fairlearn library address?

We currently focus on two kinds of harms:

  • Allocation harms. These harms can occur when AI systems extend or withhold opportunities, resources, or information. Some of the key applications are in hiring, school admissions, and lending.

  • Quality-of-service harms. Quality of service refers to whether a system works as well for one person as it does for another, even if no opportunities, resources, or information are extended or withheld.

Can the Fairlearn library be used to detect bias in datasets?

We do not have concrete plans for this at the present time.

Can the Fairlearn library recommend ways to make my model fairer?

Right now we do not have an automated tool that would help you decide which mitigation algorithm to use. Our focus is on expanding the documentation and examples to highlight when each of the algorithms might be more applicable. Note that model training is just one step in the AI development and deployment lifecycle, and other steps, such as data gathering and curation, or monitoring and debugging of the deployed system, may be better places of intervention to improve the fairness of an AI system.

What unfairness mitigation techniques does Fairlearn support?

Please see our Mitigation section.

Which ML libraries does Fairlearn support?

We have generally followed conventions of scikit-learn. However, our mitigation algorithms can be used to augment any ML algorithms that provide (or can be wrapped to provide) fit() and predict() methods. Also, any classification or regression algorithm can be evaluated using our metrics.

Does Fairlearn support multi-class classification?

On the assessment side, Fairlearn’s metrics.MetricFrame can be used with any metrics for supervised learning, including those for multi-class classification. For example, it is possible to pass sklearn.metrics.accuracy_score() or sklearn.metrics.confusion_matrix() as the metric functions, and supply multi-class data for y_true and y_pred. There are ongoing discussions within the community to add more extensive support to Fairlearn’s assessment capabilities. If you have thoughts feel free to add them to the discussion.

On the mitigation side, our algorithms reductions.ExponentiatedGradient and reductions.GridSearch support bounded group loss constraints, which are applicable to any supervised learning setting, including multi-class classification.

Does Fairlearn support multiple and non-binary sensitive features?

Fairlearn’s assessment capabilities support sensitive features with more than two values as well as multiple sensitive features. Our user guide has examples for both of these cases. The mitigation techniques all support mitigation with non-binary and multiple sensitive features as well. For a full list of techniques please refer to the user guide section on mitigation.

Does Fairlearn work for image and text data?

We have not (yet) looked at using Fairlearn on image or text data. However, so long as the image or text classifier provide fit() and predict() methods as required by Fairlearn, it should be possible to use them with Fairlearn mitigation algorithms. Also, any classification or regression algorithm can be evaluated using our metrics (regardless of the data it is operating on).

Is Fairlearn available in languages other than Python?

For the moment, we only support Python >= 3.6

Can I contribute to Fairlearn?

Absolutely! Please see our contributor guide to see how. We welcome all contributions!

What is the relationship between Fairlearn and Microsoft?

Fairlearn has grown from a project at Microsoft Research in New York City.