ETRI Successfully Concludes ‘AI That Understands People’ Paper Competition

[Financial News] The Electronics and Telecommunications Research Institute (ETRI) said on the 18th that it had successfully concluded the 5th ETRI Human Understanding AI Paper Competition, which it held to promote AI research into understanding people’s states and everyday contexts using daily-life data.
A total of 409 teams, comprising 864 people, took part in the competition. Of these, 22 teams presented research papers at the international academic conference ICTC 2026, and ETRI selected five outstanding research teams after evaluating factors including the originality and technical quality of their work.
The competition was ran for six months starting last April, with support from the Ministry of Science and ICT and the National Research Council of Science and Technology (NST), and was operated by DACON.
This year’s competition focused on “Sleep, Emotion and Stress Recognition and Inference Using Lifelog Data.” Lifelog data records activities, physical conditions and other information in everyday life.
Participants used the data to develop AI models that predicted sleep quality, fatigue and stress before going to bed, as well as total sleep time, sleep efficiency, the time it takes to fall asleep and the amount of time spent awake during sleep. In particular, the competition went beyond comparing AI models’ predictive performance, taking a research-focused approach that also assessed the academic value of the research, the completeness of the papers and the reproducibility of the findings.
After submitting their predictions on the competition website and checking their models’ performance, participants wrote up their methods and findings in papers and submitted them to the ETRI Human Understanding AI Paper Competition track at ICTC 2026. Teams whose papers passed review then submitted their program code and model documentation for verification that other researchers could reproduce the results using the same methods.
One team received the Minister of Science and ICT Award and a cash prize for the Grand Prize, while one team each received the Electronics and Telecommunications Research Institute President’s Award and a cash prize for the Best Award and the Excellence Award, and two teams received the same award and a cash prize for the Encouragement Award. The Grand Prize went to the “Meolwang” team from Korea University (Kim Gyu-bin, Park Chan-ho, Lee Gyu-ho and Kim Eun-tae). The team proposed a method to improve prediction accuracy for seven indicators related to sleep, fatigue and stress by analyzing each person’s usual state together with their recent activity records.
ETRI has held the competition annually since 2022 to make data compiled during the research available to researchers in Korea and abroad and to encourage the development of new AI algorithms and academic research using the data.

[email protected] Yeon Ji-an Reporter