The 37th Annual Conference on the Japanese Society for Artificial Intelligence, one of the largest academic conference in the field of artificial intelligence in Japan, was held in Kumamoto, Japan, June 6-9, 2023, and the presentation “Explainable Data Bias Mitigation” by Kenji Suzuki received the Annual Conference Award.
Award Name
The Japanese Society for Artificial Intelligence, Annual Conference Award
Awarded Paper
Explainable Data Bias Mitigation
Paper Abstract
This study proposes an explainable fairness method that not only mitigates data bias but also allows humans to understand the reasons for it. Machine learning algorithms require fairness, accountability, and transparency for high-risk use cases, such as hiring, lending, and other decision-making. Differences in AI predictions based on sensitive attributes such as gender, race, and age are an issue of group fairness. Various methods have been proposed for bias mitigation in AI, but conventional methods have the problem that humans cannot intuitively understand the criteria used for data bias mitigation. Therefore, we proposed a bias mitigation method using explainable AI and showed that humans can understand bias mitigation.
Winner’s Comment
I am honored to receive the prestigious Annual Conference Award of the Japanese Society for Artificial Intelligence. This research proposed an explainable bias mitigation method that not only mitigates data bias but also allows humans to understand the reasons for the bias mitigation. This research will contribute to fairness, accountability, and transparency in AI utilization. I want to express my gratitude to all the people involved in this research for their support. It was also very beneficial to have discussions with many people at the conference on the Japanese Society for Artificial Intelligence. Encouraged by this award, I would like to pursue research on AI ethics at a level that can be used in practice.
