BROWSE
Detail
Conference Paper
양식장 수질 데이터 증강을 위한 시계열 생성모델 비교 연구
Cited - time in 
Share 
Share
- Issue Date
- 2026-06
- Citation
- 대한전자공학회 학술 대회 (하계) 2026, pp.1-4
- Publisher
- 대한전자공학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- Aquaculture facilities increasingly depend on water-quality sensors for data-driven operation, but long-term field deployments frequently produce missing values and degraded readings, leaving scarce training data for automation. This paper compares four representative time-series generative models on a Haenam aquaculture water-quality dataset, jointly evaluating their distributional fidelity and downstream utility. PaD-TS records the lowest value on three of the four metrics across both axes, and the fidelity metrics further capture distributional flaws such as over-smoothing that the utility metric misses, confirming that both axes must be reported jointly.
- KSP Keywords
- Data-Driven, Missing values, Time series, generative model, over-smoothing, training data