적대적 생성 신경망을 활용한 가상 뇌파 데이터 생성 - 건축공간에 대한 사용자 선호도 파악을 위한 딥러닝 분류모델의 훈련지원을 위해
- Title
- 적대적 생성 신경망을 활용한 가상 뇌파 데이터 생성 - 건축공간에 대한 사용자 선호도 파악을 위한 딥러닝 분류모델의 훈련지원을 위해
- Other Titles
- Use of Generative Adversarial Networks(GANs) for EEG Data Augmentation - To Support Training Process of EEG-based Deep-Learning Classification Model for User Preferences toward Architectural Spaces -
- Author
- 전한종
- Keywords
- 적대적 생성 신경망; 뇌파; 감정 인식; 건물 평가; Generative Adversarial Networks; Electroencephalography(EEG); Affection Recognition; Building Evaluation
- Issue Date
- 2019-10
- Publisher
- 대한건축학회
- Citation
- 대한건축학회 학술발표대회 논문집, v. 39, no. 2, Page. 9 - 12
- Abstract
- It is important for architects to recognize subjective reponses of users toward architectural design alternatives in early phase of planning and design. In this regard, a model which analyses affective responses of decision-makers is strongly required. A previous study has structured Electroencephalography(EEG)-based deep-learning classification model for evaluating subjects’ emotional responses in quantitative manner in given experiment situation using EEG data. However, it is limited volume of EEG data that results in difficulty in training process of the model. In this regard, this paper aims to suggest Generative Adversarial Networks(GANs) which consists of generator for “fake” EEG data generation and discriminator for training the generator. GANs model may provide one possible way of wide adoption of the suggested model and structuring design knowledge database using EEG data especially for designing architectural spaces for children, elderly and patients those who interviews or questionnaires are hard to be conducted.
- URI
- http://www.dbpia.co.kr/journal/articleDetail?nodeId=NODE09289844https://repository.hanyang.ac.kr/handle/20.500.11754/154443
- ISSN
- 2287-5786
- Appears in Collections:
- COLLEGE OF ENGINEERING[S](공과대학) > ARCHITECTURE(건축학부) > Articles
- Files in This Item:
There are no files associated with this item.
- Export
- RIS (EndNote)
- XLS (Excel)
- XML