On the Fairness of Swarm Learning in Skin Lesion Classification

Fairness is essential for trustworthy Artificial Intelligence (AI) in healthcare. However, the existing AI model may be biased its decision marking. The bias induced by data itself, such as collecting subgroups only, can mitigated including more diversified data. Distributed and collaborative learning an approach to involve training models massive, heterogeneous, distributed sources, also known nodes. In this work, we target on examining fairness issue Swarm Learning (SL), a recent edge-computing based decentralized machine approach, which designed heterogeneous illnesses detection precision medicine. SL has achieved high performance clinical applications, but no attempt been made evaluate if improve fairness. To address problem, present empirical study comparing among single (node) training, SL, centralized training. Specifically, large public available skin lesion dataset, contains samples from various subgroups. experiments demonstrate that does not exacerbate problem compared improves both there still exists biases implementation of complex than alternative two strategies.