Injecting fairness into machine-learning models

MIT researchers have found that, if a certain type of machine learning model is trained using an unbalanced dataset, the bias that it learns is …