generative model6 [논문 리뷰] SDEs: Score-based generative modeling with stochastic differential equations [논문]Score-based generative modeling with stochastic differential equationsNeurIPS 2021Citations: 6,517[https://arxiv.org/abs/2011.13456] 해당 논문 보기 전 참고하면 좋은 postScore-based model 리뷰[개념 설명] Score-based Model [개념 설명] Score-based Model[Blog][https://yang-song.net/blog/2021/score/]blog 작성자 - SDE의 저명한 저자인 Yang SongNCSN, SDEs 등 Score-based model을 공부하기 전 이 blog를 통해 개념을 익히는 것 추천!!이 blog를 리뷰하는 post가 될 예정 .. 2025. 3. 25. [논문 리뷰] VDM survey: Understanding diffusion models: A Unified Perspective (1) Diffusion model 공부를 시작할 때, 관련 논문을 먼저 보는 것보다 해당 survey논문을 보는 것이 이해하기 편했음해당 논문은 Class conditional diffusion model까지 수식적으로 잘 정리되어 있는 논문해당 논문의 DDPM의 부분만 3가지 post로 나누어 정리할 예정(1) - Intro, ELBO, VAE, HVAE (2) - Diffusion model(VDM), Maximizing ELBO 2가지[논문 리뷰] VDM survey: Understanding diffusion models: A Unified Perspective (2) [논문 리뷰] VDM survey: Understanding diffusion models: A Unified Perspective (.. 2025. 3. 16. [논문 리뷰] StyleGAN: A Style-Based Generator Architecture for Generative Adversarial Networks PGGANProgressive growing of GANs for improved quality, stability, and variationICLR 2018Citations: 8796이후 GAN논문들에는 ProGAN, PGAN으로 명명Generator 부분만 변경된 model ⇒ StyleGANhttps://arxiv.org/abs/1710.10196Progressive growing of GANSPGGAN - low-resolution부터 시작하여 학습을 진행하며, 점차적으로 high-resolution의 레이어를 추가해가며 이미지를 생성하는 GAN 모델초기 단계 - low-resolution으로 부터 Large-scale structure 학습점차 finer scale detail 학습Fade-in.. 2025. 3. 13. [개념 설명] Inception scores(IS), Fréchet Inception Distance(FID), Negative Log Likelihoods(NLL) Generative model 논문에서 많이 나오는 평가 지표들에 대해 알아볼 예정Inception scores(IS)Fréchet Inception Distance(FID)Negative Log Likelihoods(NLL)Inception scores(IS)Fidelity(real한 image 생성되는지)Diversity(다양한 image 생성되는지)pre-trained inception network사용 ⇒ CNN based model⇒ 높을수록 생성된 Image 품질 & 다양성 good$IS(G) = exp(E_{x\sim p_g}[D_{KL}((p(y|x)||p(y))])$G: generator model // $x\sim p_g$: G에서 생성된 imagep(y|x): Inception net.. 2025. 3. 13. [논문 리뷰] Score-MRI: Score-based Diffusion Models for Accelerated MRI https://arxiv.org/abs/2110.05243Medical Image Analysis 2022Citations: 371 Abstract[Score-based diffusion model]learned score ft를 prior로 활용continuous time-dependent score ft with denoising score matcinginverse problems for accelerated MRIInference(reconstruction) - numerical SDE solver & data consistency step[Proposed method]In training, magnitude images 필요 but complex-valued data로 복원 가능sub-sampl.. 2025. 3. 10. [논문 리뷰] X-Diffusion: Generating Detailed 3D MRI Volumes From a Single Image Using Cross-Sectional Diffusion Models https://arxiv.org/abs/2404.196042025 ICLR submittedCitations: 2 Summary[Motivation]Whole-body MRI의 평균 소요 시간 50분Goal: Build a cost-effective AI-based MRI reconstruction[Related Work]Score-MRI (2022)- kaist 예종철 교수님Score-based Diffusion Models for Accelerated MRI[논문 리뷰] Score-MRI: Score-based Diffusion Models for Accelerated MRI [논문 리뷰] Score-MRI: Score-based Diffusion Models for Accelerated MRIhtt.. 2025. 3. 10. 이전 1 다음
