Big Data Analysis: Application to Environmental Research and Service(Ⅵ)
- Author Hong, Hanwoom
- Researchers
-
Date
2022-12-31
Ⅰ. Background and Aims of Research
oStudy purpose: Set up environmental policy monitoring system (tentative name)
ㅇPurpose is periodic repetition of understanding the demand for environmental policy, evaluating timeliness of policy, and evaluating effectiveness of policy
ㅇConsists of integrated environmental pollution prediction algorithm, real-time environmental text analysis algorithm, and query-based database
oStudy objectives for 2022: Improve the environmental policy monitoring system components (tentative name)
ㅇ Pollution degree prediction
- Improve explainability of air and water pollution degree prediction
ㅇ Query-based database
-Set up an issue board for the Korea Sustainable Development Goals (K-SDGs)
ㅇ Real-time environmental text analysis algorithm
- Output of environmental big data platform
Ⅱ.Explainable artificial intelligence-based new fine dust prediction module development
o Study purpose
ㅇStudy purpose: Introduce artificial intelligence technique to explain the fine dust prediction
- Assure artificial intelligence prediction model transparency and reliability
o Study method
ㅇ Collect data from Air Korea and Korea Meteorological Administration
ㅇ Preprocess data for missing values using spatial interpolation
ㅇ Derive attention score and shapely value using Vision Transformer (ViT)
oSet up explainable artificial intelligence prediction model while maintaining high short-term prediction performance
ㅇ Combine ViT to Interpolated Convolutional Neural Network (ICNN)
ㅇ Derive the impact of a specific patch on the results through attention score
ㅇ Understand t
oStudy purpose: Set up environmental policy monitoring system (tentative name)
ㅇPurpose is periodic repetition of understanding the demand for environmental policy, evaluating timeliness of policy, and evaluating effectiveness of policy
ㅇConsists of integrated environmental pollution prediction algorithm, real-time environmental text analysis algorithm, and query-based database
oStudy objectives for 2022: Improve the environmental policy monitoring system components (tentative name)
ㅇ Pollution degree prediction
- Improve explainability of air and water pollution degree prediction
ㅇ Query-based database
-Set up an issue board for the Korea Sustainable Development Goals (K-SDGs)
ㅇ Real-time environmental text analysis algorithm
- Output of environmental big data platform
Ⅱ.Explainable artificial intelligence-based new fine dust prediction module development
o Study purpose
ㅇStudy purpose: Introduce artificial intelligence technique to explain the fine dust prediction
- Assure artificial intelligence prediction model transparency and reliability
o Study method
ㅇ Collect data from Air Korea and Korea Meteorological Administration
ㅇ Preprocess data for missing values using spatial interpolation
ㅇ Derive attention score and shapely value using Vision Transformer (ViT)
oSet up explainable artificial intelligence prediction model while maintaining high short-term prediction performance
ㅇ Combine ViT to Interpolated Convolutional Neural Network (ICNN)
ㅇ Derive the impact of a specific patch on the results through attention score
ㅇ Understand t