ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술대회 Explanation for Building Energy Prediction
Cited 7 time in scopus Download 7 time Share share facebook twitter linkedin kakaostory
저자
김말희, 전종암, 송유진, 표철식
발행일
202010
출처
International Conference on Information and Communication Technology Convergence (ICTC) 2020, pp.1168-1170
DOI
https://dx.doi.org/10.1109/ICTC49870.2020.9289340
초록
Explainable AI is a technology field that has recently emerged. Currently, deep learning technology shows high accuracy in many areas. However, despite its high accuracy, there are areas where technology cannot be utilized without sufficient explanation of the results. In medicine, credit rating, and criminal fields, it is difficult to apply technology in practice without sufficient explanation of the results. This paper gives an explanation of the results of the energy demand prediction model by using feature importance and attention mechanism. The model used is seq2seq model, and the dataset used the data set generated by energyPlus?꽓 based on the actual data.
KSP 제안 키워드
Attention mechanism, Building energy prediction, Credit rating, Data sets, Feature Importance, High accuracy, deep learning(DL), energy demand prediction, learning technology, prediction model