Major Research Achievements

Professor Lee Byung-joo's research team presented three papers at CHI'…

페이지 정보


작성자 최고관리자

댓글 0 comment 조회 286 view 작성일 2022-03-02 15:43


Professor Lee Byung-joo's research team will present three papers at CHI'2022, the most prestigious society in the field of human-computer interaction.  

[Thesis 1] “Quantifying Proactive and Reactive Button Input, Hyunchul Kim, Kasper Hornbaek, and Byungjoo Lee (corresponding author)”

[Thesis 2] “How AI-based Traning Changed Performance of Professional Go Players, Jimoon Kang, June-Seop Yoon (co-first author), and Byungjoo Lee (corresponding author)”

[Thesis 3] “Speeding up Inference with User Simulators through Policy Modulation, Heeseung Moon, Seungwon Do, Wonjae Kim, Minsuk Chang (co-corresponding author), Jiwon Seo (co-corresponding author), and Byungjoo Lee (co-corresponding author)”

Paper 1, a joint study with the University of Copenhagen, KAIST, proposed and validated a new analysis technique that quantifies how responsive or preemptive user input was performed with only video recording computer screens and user input signals. This study was also applied for two international patents in the United States and Japan in December 2021.

Paper 2 analyzed how the notation pattern of professional Go players changed after AlphaGo appeared. After AlphaGo's appearance, Go training based on artificial intelligence programs has become common, and the notes of Go drivers have gradually become similar to those recommended by artificial intelligence. Furthermore, as the time spent on artificial intelligence-based training rather than new and creative strategies became a major factor in determining the performance of Go players, artificial intelligence-based training was found to have benefited senior players more than junior players.  

Paper 3 is a joint study with the Department of Computer Science, the Department of Global Convergence Engineering, ETRI, and Naver. Using a point-and-click simulator developed by Professor Lee Byung-joo in 2021, we proposed a more efficient method to determine the free parameter value of the simulator from actual human point-and-click behavior data. Through this technique, the parameter estimation process, which previously took several days or more, can be shortened to within several hours.

The three studies will be presented at the CHI'2022 conference, which begins on April 30 in the United States.


등록된 댓글이 없습니다.