POSTECH: Graduate Students Jinhyeok Park and Dain Kim Win at International Algorithm Contest

While artificial intelligence is utilized in analyzing judicial precedents as well as in hiring processes, AI bias on gender, nationality and race remains one of the biggest issues. Recently, two POSTECH graduate students have won a prize at a contest designed to tackle these specific AI analysis issues, called the CIKM 2022 Recommendation System Evaluation Challenge held by one of the most prestigious international conferences in the data mining field.

Jinhyeok Park and Dain Kim (advisor: Professor Dongwoo Kim) in the master’s program at POSTECH’s Graduate School of Artificial Intelligence have won the second place and the best paper award at the Recommendation System Evaluation Challenge sponsored by the International Conference on Information & Knowledge Management (CIKM) and hosted by an open-source community, RecList.

The task of the contest was to develop an algorithm that provides an impartial recommendation to users by removing various biases in the given data. The two students developed and proposed an algorithm after efficiently removing the bias with variational autoencoder technology.

Jinhyeok Park and Dain Kim will further develop the technology published in the paper and stated, “It is already well-known that artificial intelligence has a bias against specific groups.” They added, “If an issue of bias is resolved, we believe artificial intelligence will be utilized in broader fields.”

The competition was fierce with 50 teams from 14 countries that participated in the contest over a two-month period from August to October of 2022.