Ethan's Winery

  • 홈
  • 방명록

GAN 1

Mode collapse

Mode Collapse : GAN 에서 학습시키려는 모델이 실제 데이터를 커버하지 못하고 다양성을 잃어버리는 현상 위 그림에서 BSA 의 예시를 보게 되면 모두 다 비슷한 이미지를 생성한다는 것을 관찰할 수 있다. 이런 문제는 Discriminator 와 Generator 학습의 불균형으로 Generator 가 계속 같은 종류의 이미지를 생성하려고 하면서 발생하게 된다. 참고문헌 [1] Li, Yijun, et al. "Few-shot image generation with elastic weight consolidation." arXiv preprint arXiv:2012.02780 (2020).

AI/Deep Learning 2022.01.26
이전
1
다음
더보기
프로필사진

Ethan's Winery

이성훈 Ethan

  • 분류 전체보기
    • AI
      • Deep Learning
      • Metrics
    • Computer Vision and Graphic..
      • CV Definitions
      • 3D CV Lectures
      • Multiple View Geometry
    • Paper Review
      • [Paper Survey]
      • Object Detection, Pose esti..
      • Human Mesh Pose
      • GAN, Inversion
      • Diffusion, VAE
      • 3D Reconstruction (3DGS, NE..
      • Continual Learning (CL)
      • Privacy Protection (UL, Ano..
      • Out-of-Distribution Detecti..
      • Novel Class Discovery (NCD)
      • Multi-Modal (MM)
      • Self-Supervised Learning (S..
      • Scene Change Detection (SCD..
      • Video Understanding
    • Code & Framework
      • Paper Open Source Codes
      • Python Numpy
      • Pytorch
      • Linux Conda
      • HTML & CSS
      • C++
      • NVIDIA CUDA
      • NVIDIA Omniverse
      • Git SVN
      • ONNX
      • Docker
      • ETC
    • Side Project
      • NVIDIA Jetson Orin Nano
    • Computer Science
      • Data Structure, Algorithm
      • Comp Arch, OS
      • Database
    • Mathematics
      • Linear Algebra
      • Probability Statistics
    • ETC
      • 코딩테스트
      • 잡동사니
      • 학회
250x250

Tag

용어, 딥러닝, Continual Learning, dl, incremental learning, GAN, image classification, fewshot,

최근글과 인기글

  • 최근글
  • 인기글

공지사항

Archives

Calendar

«   2025/10   »
일 월 화 수 목 금 토
1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31

방문자수Total

  • Today :
  • Yesterday :

Copyright © Kakao Corp. All rights reserved.

티스토리툴바