An Unsupervised learning model for deformable medical image registration¶ 1. Introduction¶1.1 deformable registration과 문제점¶ deformable registration fundamental task in a variety of medical imaging studies topic of active research for decades N차원 영상들의 pair에 대해서 dense하고 non-linear한 일치(correpondance)가 이루어짐 voxel: pixel의 3차원 버전 대부분의 registraion(정합) 방법들은 비슷한 외형으로 voxel이 정렬된(align) 각 volume pair에 대해 o..
Untitled6 Ambient GAN¶ 1 Introduction¶1-1생성모델¶ 생성 모델은 대규모 데이터 세트의 구조를 간결하게 나타낼 수있는 강력한 도구 암시 적 생성 모델은 확률 분포로부터 샘플을 생성하기위한 확률 과정만을 지정하는 메커니즘 확률 분포에 대한 명시적 parameterization 필요 없음 명시적 생성 모델 vs 암시적 생성 모델 명시적 모델 $p(x)->probability,\ x:\ sample$ x가 나올 확률을 계산 해줌 암시적 모델 $f(z) -> x,\ x:\ sample,\ z:\ noise$ 랜덤 input z에 대해 x를 생성 1-2 GAN¶ GAN: 암시적 모델에서 뛰어난 성능 보여줌 geneartor: low-dimensional input -> high-di..
Untitled8 Progressive Growing GAN1. Introduction¶1-1 생성 모델의 종류¶ 생성모델은 다양한 활용 예시를 가지고 있음 음성합성 image to imgae translation image inpainting 요새 많이 쓰이는 3가지 모델 (각자 장점, 단점 있음) GAN 장점 sharp images 단점 작은 레졸류션에서만 가능 limited variation 학습 불안정 despite recent progress VAE 장점 VAEs are easy to train 단점 요즘 연구가 보완하긴 했지만 흐릿한 결과 autoregressive model 장점 sharp image 단점 slow to evaluate latent representation 없음 direct..
Self-attention-GAN Self-attention GAN¶ 1. Introduction¶1.1 GAN 소개 및 단점¶ GAN은 성공적이였으나 일부 클래스에 대해 문제점이 있음 GAN은 텍스처 위주 클래스(바다, 하늘 등)는 잘 생성 그러나 geometry가 중요한 클래스 (동물 등) 는 한계가 있음 geometric or structural patterns 표현은 잘 못함 생성할 때 동물의 texture만은 실제같음 한계에 대한 이유 GAN에서는 이미지의 다양한 위치에 다한 dependency를 모델링하기 위해서 convolution에 많이 의지함 convolution의 receptive field의 범위는 local로 제한됨. 따라서 거리가 먼 region 사이의 dependency 모델링은..
- Total
- Today
- Yesterday
- regularization
- progressive
- stochastic neighborhood embedding
- 머신러닝
- self attention gan
- GAN
- 고차원
- self attention
- 한글
- attention gan
- progressive growing gan
- No Free Lunch
- 배치노말라이제이션
- 배치 정규화
- 데이터 시각화
- high dimension
- Deep Learning
- self-attention
- Ambient
- 정리
- t-SNE
- internal covariate shift
- batch normalization
- growing
- 요약
- 한국어
- SNE
- PGGAN
- Ambient GAN
- 설명
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 |