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 |
Tags
- WBP
- CNN
- ddpm
- ue5.4
- Diffusion
- deep learning
- Few-shot generation
- WinAPI
- Generative Model
- NLP
- UE5
- GAN
- RNN
- multimodal
- userwidget
- Unreal Engine
- 언리얼엔진
- Font Generation
- Stat110
- 생성모델
- cv
- 딥러닝
- motion matching
- 디퓨전모델
- animation retargeting
- dl
- 모션매칭
- 오블완
- BERT
- 폰트생성
Archives
- Today
- Total
목록Score-based model (1)
Deeper Learning

Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon, Ben Poole, (2020.11) [Stanford, Google] Why score function? 생성모델을 학습한다는 것은 데이터 분포 pθ(x) 를 근사하는 것과 같다. likelihood based 모델에서는 직접적으로 pmf 또는 pdf를 모델링하는데 아래 식처럼 학습가능한 parameter set θ 로 parameterize된 Real-valued function fθ(x) 를 사용하여 pdf를 정의할 수 있다. $$ p_\theta(x) = \frac{e^{-f_\theta}(x)}{Z_\th..
AI/Deep Learning
2023. 5. 27. 21:24