Bias in Machine Learning Software: Why? How? What to do?
This addresses bias affecting social groups in critical applications like criminal sentencing and hiring, though it appears incremental as it builds on existing fairness improvement approaches.
The paper tackles bias in machine learning systems by postulating that root causes are in data selection and labeling decisions, and presents Fair-SMOTE which removes biased labels and rebalances distributions. The method reduces bias as effectively as prior approaches while achieving higher recall and F1 scores than other state-of-the-art fairness algorithms.
Increasingly, software is making autonomous decisions in case of criminal sentencing, approving credit cards, hiring employees, and so on. Some of these decisions show bias and adversely affect certain social groups (e.g. those defined by sex, race, age, marital status). Many prior works on bias mitigation take the following form: change the data or learners in multiple ways, then see if any of that improves fairness. Perhaps a better approach is to postulate root causes of bias and then applying some resolution strategy. This paper postulates that the root causes of bias are the prior decisions that affect- (a) what data was selected and (b) the labels assigned to those examples. Our Fair-SMOTE algorithm removes biased labels; and rebalances internal distributions such that based on sensitive attribute, examples are equal in both positive and negative classes. On testing, it was seen that this method was just as effective at reducing bias as prior approaches. Further, models generated via Fair-SMOTE achieve higher performance (measured in terms of recall and F1) than other state-of-the-art fairness improvement algorithms. To the best of our knowledge, measured in terms of number of analyzed learners and datasets, this study is one of the largest studies on bias mitigation yet presented in the literature.