A New Reinforcement Learning-Based Framework for Unbiased Autonomous Software Systems

From the newsfeeds, the products and services advertised to us, job screening, risk analysis, facial recognition and to the results we get through search engines, human-curated algorithms sitting behind the scenes, are making these decisions. These algorithms sometimes display the choices of those who authored them. Algorithmic bias has always been a minor issue since the advent of computer software, not until now that computer algorithms are deeply rooted in our daily lives through smart devices, intelligent software solutions and Autonomous System. We are at the edge of leaving our critical decisions in the hands of these intelligent creations of ours. Whereas, the biases in the algorithms used to develop them and the biases in the data they were trained with are obviously still in existence. This research is aimed at adapting the existing bias detection mechanism to ensure fairness in the decision-making process of autonomous software systems. The final solution is in the form of a software module which provides access for auditing decision process of machine learning powered autonomous software systems. This, in turn, ensures fairness in the decision process of autonomous software systems.