Fairness in Reinforcement Learning
This work tackles the problem of ensuring fairness in reinforcement learning applications, which is crucial for real-world deployments impacting multiple stakeholders, though it appears incremental as it adapts existing fairness concepts to this context.
The paper addresses the lack of fairness considerations in reinforcement learning systems, such as decision support and autonomous systems, which can lead to unfair outcomes for users or stakeholders, and proposes using social welfare functions to encode fairness as a solution.
Decision support systems (e.g., for ecological conservation) and autonomous systems (e.g., adaptive controllers in smart cities) start to be deployed in real applications. Although their operations often impact many users or stakeholders, no fairness consideration is generally taken into account in their design, which could lead to completely unfair outcomes for some users or stakeholders. To tackle this issue, we advocate for the use of social welfare functions that encode fairness and present this general novel problem in the context of (deep) reinforcement learning, although it could possibly be extended to other machine learning tasks.