Fairness Aware Swarm-based Machine Learning for Data Streams

Machine learning has been widely applied to extract insights from streaming data. However, ethical issues such as fairness have emerged related to these decision-support systems. Feature engineering methods have shown potential in representing and learning of fairness learning. However, these techniques have not been applied to streaming data. In this paper, we proposed a fairness-aware swarm-based machine learning for streaming data. The novelty of this algorithm is in the utilisation of two swarms, one for classification by building a network of prototypes and one for discrimination mitigating using feature weighting. Experiments with well-known datasets in fairness learning show that the proposed methods can improve fairness while maintaining the classification performance.