Sensor-Based Multi-Label Dataset Analysis Challenge: Predicting Sleep Quality and Emotional States in Daily Life

We have collected a multimodal lifelog dataset of daily human behavior and sleep activities from smartphones, smartwatches, sleep sensors, and self-survey records. As part of the research on AI techniques to understand human daily life using this dataset, we held the third Human Understanding AI Paper Challenge in 2024, a data analysis and research paper competition, attracting over 220 participants. This paper describes the background and dataset of this year's competition, and presents an example of data utilization and a machine learning baseline model. The final performance results of the participating teams and the top seven winners will be announced at the ICTC 2024 Workshop on ETRI Human Understanding AI Paper Challenge.