Publications
Detailed Information
Information-based Boundary Equilibrium Generative Adversarial Networks with Disentangled Representation Learning : 데이터의 특징을 해석 가능하게 학습하여 양질의 다양한 이미지를 생성하는 개선된 알고리즘에 관한 연구
DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | 이재욱 | - |
dc.contributor.author | 하정훈 | - |
dc.date.accessioned | 2018-05-29T03:20:44Z | - |
dc.date.available | 2018-05-29T03:20:44Z | - |
dc.date.issued | 2018-02 | - |
dc.identifier.other | 000000149370 | - |
dc.identifier.uri | https://hdl.handle.net/10371/141439 | - |
dc.description | 학위논문 (석사)-- 서울대학교 대학원 : 공과대학 산업공학과, 2018. 2. 이재욱. | - |
dc.description.abstract | This paper describes a new image generation algorithm, developed from BEGAN. With an information-theoretic extension to BEGAN, this new algorithm is able to learn disentangled representations from image data sets. The information-theoretic extension was inspired by InfoGAN, a generative adversarial network that maximizes the mutual information between small subset of the latent variables and the observation. By using the proposed method of this paper, one can manipulate the generated images as desired by controlling the latent codes of input variables. Also the visual qualities of produced images are highly maintained. | - |
dc.description.tableofcontents | Chapter 1 Introduction 1
1.1 Motivation of the Research 1 1.2 Aims of the Research 2 Chapter 2 Related Works 3 2.1 BEGAN 3 2.1.1 Wasserstein distance lower bound 4 2.1.2 GAN objective 5 2.1.3 Equilibrium concept 6 2.1.4 Boundary Equilibrium Generative Adversarial Network 6 2.2 InfoGAN 7 2.2.1 Unsupervised learning 8 2.2.2 Works related to InfoGAN 8 2.2.3 Mutual Information for latent codes 9 2.2.4 Lower bound of mutual information 11 Chapter 3 Proposed Method 13 3.1 Model architecture 13 3.2 Objective of proposed model 15 Chapter 4 Experiments 17 4.1 Data description 17 4.1.1 CelebA 17 4.1.2 LSUN 17 4.2 Disentangled Representation 18 4.2.1 Results with CelebA 18 4.2.2 Results with LSUN 20 Chapter 5 Conclusion 23 5.1 Summary 23 5.2 Future Work 23 Bibliography 25 국문초록 29 | - |
dc.format | application/pdf | - |
dc.format.extent | 8848917 bytes | - |
dc.format.medium | application/pdf | - |
dc.language.iso | en | - |
dc.publisher | 서울대학교 대학원 | - |
dc.subject | Generative Adversarial Networks | - |
dc.subject | Disentangled representation learning | - |
dc.subject | Mutual information | - |
dc.subject.ddc | 670.42 | - |
dc.title | Information-based Boundary Equilibrium Generative Adversarial Networks with Disentangled Representation Learning | - |
dc.title.alternative | 데이터의 특징을 해석 가능하게 학습하여 양질의 다양한 이미지를 생성하는 개선된 알고리즘에 관한 연구 | - |
dc.type | Thesis | - |
dc.description.degree | Master | - |
dc.contributor.affiliation | 공과대학 산업공학과 | - |
dc.date.awarded | 2018-02 | - |
- Appears in Collections:
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.