DOI QR코드

DOI QR Code

신제품 개발을 위한 GAN 기반 생성모델 성능 비교

Performance Comparisons of GAN-Based Generative Models for New Product Development

  • 이동훈 (경북대학교 인공지능학과) ;
  • 이세훈 (경북대학교 인공지능학과) ;
  • 강재모 (경북대학교 인공지능학과)
  • 투고 : 2022.10.11
  • 심사 : 2022.11.04
  • 발행 : 2022.11.30

초록

최근 빠른 유행의 변화 속에서 디자인의 변화는 패션기업의 매출에 큰 영향을 미치기 때문에 기업들은 신제품디자인 선택에 신중할 수밖에 없다. 최근 인공지능 분야의 발달에 따라 패션시장에서도 소비자들의 선호도를 높이기 위해 다양한 기계학습을 많이 활용하고 있다. 우리는 선호도와 같은 추상적인 개념을 수치화함으로써 신제품 개발에 신뢰성을 높이는 부분에 기여하고자 한다. 이를 위해 3가지 적대적 생성 신경망(Generative adversial netwrok, GAN)을 통하여 기존에 없는 새로운 이미지를 생성하고, 미리 훈련된 합성곱 신경망(Convolution neural networkm, CNN)을 이용하여 선호도라는 추상적인 개념을 수치화시켜 비교하였다. 심층 컨볼루션 적대적 생성 신경망(Deep convolutional generative adversial netwrok, DCGAN), 점진적 성장 적대적 생성 신경망(Progressive growing generative adversial netwrok, PGGAN), 이중 판별기 적대적 생성 신경망(Dual Discriminator generative adversial netwrok, D2GAN)의 3가지 방법을 통해 새로운 이미지를 생성하였고, 판매량이 높았던 제품으로 훈련된 합성곱 신경망으로 유사도를 비교, 측정하였다. 측정된 유사도의 정도를 선호도로 간주하였으며 실험 결과 D2GAN이 DCGAN, PGGAN에 비해 상대적으로 높은 유사도를 보여주었다.

Amid the recent rapid trend change, the change in design has a great impact on the sales of fashion companies, so it is inevitable to be careful in choosing new designs. With the recent development of the artificial intelligence field, various machine learning is being used a lot in the fashion market to increase consumers' preferences. To contribute to increasing reliability in the development of new products by quantifying abstract concepts such as preferences, we generate new images that do not exist through three adversarial generative neural networks (GANs) and numerically compare abstract concepts of preferences using pre-trained convolution neural networks (CNNs). Deep convolutional generative adversarial networks (DCGAN), Progressive growing adversarial networks (PGGAN), and Dual Discriminator generative adversarial networks (DANs), which were trained to produce comparative, high-level, and high-level images. The degree of similarity measured was considered as a preference, and the experimental results showed that D2GAN showed a relatively high similarity compared to DCGAN and PGGAN.

키워드

과제정보

이 논문은 2022년도 4단계 두뇌한국21 사업(4단계 BK21 사업)에 의하여 지원되었음. 본 논문은 2022년도 정부(교육부)의 재원으로 한국연구재단의 지원을 받아 수행된 기초연구사업(No. 2020R1I1A3073651)의 지원을 받아 작성되었음.

참고문헌

  1. Seyoon Jang and Ha Youn Kim, "Deep Learning for Classification of High-End Fashion Brand Sensibility", Journal of the Korean Society of Clothing and Textiles, pp 165-181, (2022). https://doi.org/10.5850/JKSCT.2012.36.2.165
  2. Park Hakgwon, Young-Hwan Lim and Bin Lin, "Analysis Product Recommendation Service Using Image-Based AI Skin Color Detecting Technology", The Journal of the Convergence on Culture Technology(JCCT), 8(3), pp 501-506. (2022). https://doi.org/10.17703/JCCT.2022.8.3.501
  3. Hyun-moon Park, Byung-chan Jeon and Daehyun Ryu, "A Study for Context-Awareness based on Multi-Sensor in the Smart-Clothing", The Journal of the Institute of Internet, Broadcasting and Communication, 13(3), pp 71-78, (2013). https://doi.org/10.7236/JIIBC.2013.13.3.71
  4. Bongkyu Lee, "A Study on the Analysis of Structural Textures using CNN (Convolution Neural Network)", The Journal of the Institute of Internet, Broadcasting and Communication, 20(4), pp 201-205, 2020). https://doi.org/10.7236/JIIBC.2020.20.4.201
  5. Hee-young Kim, Sun-mi Jung, Woo-suk Kim, Gi-hwan Ryu and Hyeon-kon Son, "Study on Implementation of Restaurant Recommendation System based on Deep Learning-based Consumer Data", The Journal of the Convergence on Culture Technology (JCCT), 7(2), pp 437-442, (2021). https://doi.org/10.17703/JCCT.2021.7.2.437
  6. Yong Jun Kim and Jung Min Park, "A study of creative human judgment through the application of machine learning algorithms and feature selection algorithms", The International Journal of Advanced Smart Convergence, 11(2), pp 38-43, (2022). https://doi.org/10.7236/IJASC.2022.11.2.38
  7. Naejoung Kwak, Doyun Kim, Minho kim, Jongseo kim, Sangha Myung, Youngbin Yoon and Jihye Choi, "Cody Recommendation System Using Deep Learning and User Preferences", International Journal of Advanced Culture Technology(IJACT), 7(4), pp 321-326, (2019). https://doi.org/10.17703/IJACT.2019.7.4.321
  8. Byungmin Kim, Saud Alguwaizini and Kyungsook Han, "A recommendation method based on personal preferences regarding the price, rating and selling of products", The Journal of Korea Information Processing Society(KIPS), 21(2), pp 1042-1045, (2014).
  9. Son Min Jae, Jung Seung Won and Hwang Een Jun, "A Deep Learning Based Over-Sampling Scheme for Imbalanced Data Classification", The Journal of Korea Information Processing Society (KIPS), 8(2), pp 311-316, (2019).
  10. Jiha Kim and Hyunhee Park, "Validation and eXplainable GAN(VX-GAN) model for GAN Data augmentation", Myongji Univ, Dept. Information and Communication Engineering, (2021).
  11. Han-Kyu Lee, Cheol-Hwan Yoo, Yong-Goo Shin, Se-Ho Lim and Sung-Jea Ko, "Semantic Segmentation for Fashion Items using Convolutional Neural Network", The Institute of Electronics and Information Engineers, pp 512-515, (2017).