▲ From left: Professor Park Soo-yong (supervisor), Department of Computer Science & Engineering; Lee Jun-ha (master's student, 1st semester)

The problem of bias leading to discrimination in machine learning has recently drawn attention. Meanwhile, machine learning technology is developing in the direction of federated learning, driven by improved device performance and heightened awareness of data privacy. This paper presents Seldonian BlockFlow, a methodology that allows people to set criteria for an 'appropriate algorithm' in federated learning and uses blockchain smart contracts to verify and guarantee that those criteria have been properly applied. The significance of the research lies in showing that social criteria for 'appropriate artificial intelligence' can be introduced and implemented in decentralized AI training environments.

The Korean Institute of Information Scientists and Engineers' Korea Computer Congress (KCC2021) is an information science and technology conference at which the latest research results across every area of information technology — artificial intelligence, computer systems, information security and high-assurance computing and more — are presented.