지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술저널
- 저자정보
- 저널정보
- 대한산업공학회 Industrial Engineering & Management Systems Industrial Engineering & Management Systems Vol.20 No.3
- 발행연도
- 2021.9
- 수록면
- 384 - 397 (14page)
- DOI
- 10.7232/iems.2021.20.3.384
이용수
초록· 키워드
오류제보하기
Data augmentation methods are used as a technique to improve generalization by increasing the number of training data in image classification. However, most of these methods are not a data driven algorithm, the degree of improvement of generalization ability by performing these data augmentation methods differs between the domains of image data for training. Generative models are researched to use for augmenting data recently. In particular, Generative Adversarial Networks (GANs) (Goodfellow et al., 2014) that can generate clean image get attention as an excellent innovation in machine learning. As GANs extension method, there is a method called CGANs (Mirza and Osindero, 2014) that can be used for data augmentation. When enough training data for each class are not prepared for classification model, the same is true for training CGANs. In such case, CGAN generates noisy images. This makes a classification model to underfit to the original training data. Moreover, when a CGAN approximates the training data distribution, the CGAN generates new training data in the same region where training data densely exist. In such case, augmented data can’t reduce overfitting on the original training data. Therefore, our research contributes to augment data which meets these two requirements. In this study, we propose a method to generate data by the class specific GAN with small training data and selectively add generated data to the training data set that improves classification accuracy by using the entropy of the classification model. The feature of the proposed method is that it focuses on the positional relationship between data and the classification hyperplane in deep learning. In the proposed method, the entropy of the classification model is used to measure the positional relationship between the classification boundary and the data. As a result, the generalization performance is improved by adding the data around the classification boundary as new training data.
목차
ABSTRACT
1. INTRODUCTION
2. RELATED WORK
3. PRELIMINARIES
4. PROPOSAL
5. EXPERIMENT
6. CONCLUSION AND FUTURE WORKS
REFERENCES
참고문헌 (25)
참고문헌 신청
Akaike, H. / 1998 / Information theory and an exten-sion of the maximum likelihood principle / Springer :199~213
Arjovsky, M. / 2017 / Was-serstein generative adversarial networks, Proceed-ings of the 34th International Conference on Ma-chine Learning / Proceed-ings of the 34th International Conference on Ma-chine Learning, PMLR 70 :214~223
Dai, Z / Good semi supervised training that requires a bad GAN, In Advances in Neural Information Processing Systems, 6513-6523
Goodfellow, I. J. / 2014 / Generative adversarial nets / Pro-ceedings of the 27th International Conference on Neural Information Processing Systems :2672~2680
Gulrajani, I. / 2017 / Improved training of wasserstein GANs / Proceedings of the 31st Interna-tional Conference on Neural Information Pro-cessing Systems :5769~5779
Arjovsky, M. / 2017 / Was-serstein generative adversarial networks, Proceed-ings of the 34th International Conference on Ma-chine Learning / Proceed-ings of the 34th International Conference on Ma-chine Learning, PMLR 70 :214~223
Dai, Z / Good semi supervised training that requires a bad GAN, In Advances in Neural Information Processing Systems, 6513-6523
Goodfellow, I. J. / 2014 / Generative adversarial nets / Pro-ceedings of the 27th International Conference on Neural Information Processing Systems :2672~2680
Gulrajani, I. / 2017 / Improved training of wasserstein GANs / Proceedings of the 31st Interna-tional Conference on Neural Information Pro-cessing Systems :5769~5779
함께 읽어보면 좋을 논문
논문 유사도에 따라 DBpia 가 추천하는 논문입니다. 함께 보면 좋을 연관 논문을 확인해보세요!
-
영상 내 물체 검출 및 분류를 위한 소규모 데이터 확장 기법
로봇학회 논문지
2020 .06
-
Frequency-Domain Data Augmentation of Vibration Data for Fault Diagnosis using Deep Neural Networks
제어로봇시스템학회 국제학술대회 논문집
2022 .11
-
CGAN 을 이용한 생성 모델의 결과 데이터의 실효성에 관한 연구
한국통신학회 학술대회논문집
2019 .01
-
A New Image Classification Method
INTERNATIONAL CONFERENCE ON FUTURE INFORMATION & COMMUNICATION ENGINEERING
2021 .02
-
특징 벡터 다양화를 통한 cGANs의 데이터 증강 개선 연구
한국정보과학회 학술발표논문집
2020 .07
-
CNN 기반 조건 예측을 통한 cGAN 개선 방법
한국정보통신학회 종합학술대회 논문집
2023 .10
-
Experimental Analysis of Equilibrization in Binary Classification for Non-Image Imbalanced Data Using Wasserstein GAN
International journal of internet, broadcasting and communication : IJIBC
2019 .01
-
Few-shot Learning-Based Image Classification
INTERNATIONAL CONFERENCE ON FUTURE INFORMATION & COMMUNICATION ENGINEERING
2022 .01
-
표면 결함 검출을 위한 데이터 확장 및 성능분석
전기학회논문지
2018 .05
-
Data Augmentation on Limited Biometric Data Set for M2M Authentication Model Testing
한국통신학회 학술대회논문집
2021 .06
-
The Role and Effect of Data Augmentation in Generative Adversarial Nets
한국정보과학회 학술발표논문집
2021 .06
-
기계의 결함음 검출을 위한 학습 데이터 합성 및 데이터 선별법에 관한 연구
전기학회논문지
2024 .07
-
로봇 디팔레타이징 산업용 인공지능 물체 검출 학습의 데이터 비용 절감을 위한 GAN 기반 데이터 증강 기법
제어로봇시스템학회 논문지
2022 .10
-
A Novel Rar e Image Data Augmentation Method for Fish Diseases Classification
멀티미디어학회논문지
2025 .02
-
Effect of Augmentation Policies of Supervised Learning to Generative Adversarial Networks
한국정보과학회 학술발표논문집
2021 .12
-
Diffusion 및 BLIP 모델 기반 이미지 데이터 증강 기법
대한전자공학회 학술대회
2023 .06
-
벡터 양자화 증강 기법을 통한 적대적 견고 향상
대한전자공학회 학술대회
2023 .06
-
Few-shot 학습 기반 이미지 분류
한국정보통신학회 종합학술대회 논문집
2021 .05
-
Novel Image Classification Method Based on Few-Shot Learning in Monkey Species
Journal of information and communication convergence engineering
2021 .06
-
데이터 증강기법 기반의 강건한 건선 중증도 분류 연구
전기학회논문지
2022 .12
이 논문의 저자 정보
최근 본 자료
댓글(0)
0