상세 보기
초록
This study proposes a methodology to enhance the classification performance of traditional machine learning models for lifelog data analysis by leveraging Large Language Models (LLMs) for missing value imputation. The reasoning capabilities of LLMs are used throughout the imputation process, and the resulting features are fed into an ensemble-based predictor. In our evaluation, the proposed approach improved classification accuracy on lifelog data compared with conventional methods.
키워드
Lifelog; LLM; Machine Learning Ensemble; Missing Value Imputation; Sleep; Sleep Quality; Stress Level
- 제목
- Sleep Quality Prediction from Lifelog Data Using LLM-based Imputation
- 저자
- Hyun, JongYeol; Byun, YuChul; Bak, Dongkeun
- 발행일
- 2025
- 유형
- Conference Paper
- 저널명
- International Conference on ICT Convergence
- 페이지
- 1721 ~ 1726