ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술지 Improved Recurrent Generative Adversarial Networks with Regularization Techniques and a Controllable Framework
Cited 0 time in scopus Download 6 time
저자
이민혁, 태동현, 최재훈, 정호열, 석준희
발행일
202010
출처
Information Sciences, v.538, pp.428-443
ISSN
0020-0255
출판사
Elsevier
DOI
https://dx.doi.org/10.1016/j.ins.2020.05.116
협약과제
20HR4400, 심혈관질환을 위한 인공지능 주치의 기술 개발, 김승환
초록
© 2020 Elsevier Inc. Generative Adversarial Network (GAN), a deep learning framework to generate synthetic but realistic samples, has produced astonishing results for image synthesis. However, because GAN is routinely used for image datasets, regularization methods for GAN have been developed for convolutional layers. In this study, to expand these methods for time-series data, which are one of the most common data types in various real datasets, modified regularization methods are proposed for Long Short-Term Memory (LSTM)-based GANs. Specifically, the spectral normalization, hinge loss, orthogonal regularization, and the truncation trick are modified and assessed for LSTM-based GANs. Furthermore, a conditional GAN architecture called Controllable GAN (ControlGAN) is applied to LSTM-based GANs to produce the desired samples. The evaluations are conducted with sine wave data, air pollution datasets, and a medical time-series dataset obtained from intensive care units. As a result, ControlGAN with the spectral normalization on gates and cell states consistently outperforms the others, including the conventional model, called Recurrent Conditional GAN (RCGAN).
키워드
Generative adversarial network, Long short-term memory, Recurrent neural network, Sample generation, Spectral normalization
KSP 제안 키워드
Air Pollution, Common data, Data type, Deep learning framework, Hinge loss, Intensive Care, Long short-term memory, Recurrent Neural Network(RNN), Regularization methods, Regularization technique, Sine wave