Score-based Generative Models

Score-Based Generative Models is a class of generative modeling techniques within probabilistic machine learning that distinguishes itself by formulating the generation process through the estimation and inversion of score functions—gradients of the data density—to progressively denoise samples from a prior distribution back to the data manifold.

76

Publications

6K

Citations

285

Authors

93

Institutions

Publications per year

2017–2026

74

Variants

Score-based Generative ModelingScore-based Models

Authors

285

  1. SE
    Stefano Ermon

    Stanford University, United States

    8 publications · 2.8K citations

    8

    h-index

  2. YS
    Yang Song

    Siemens (China), China

    7 publications · 2.8K citations

    7

    h-index

  3. QL
    Qiegen Liu

    Nanchang University, China

    5 publications · 205 citations

    5

    h-index

  4. IM
    Il‐Chul Moon

    Korea Advanced Institute of Science and Technology, South Korea

    3 publications · 33 citations

    3

    h-index

  5. WK
    Wanmo Kang

    Korea Advanced Institute of Science and Technology, South Korea

    3 publications · 33 citations

    3

    h-index

1–5 of 285

Venues

  1. arXiv (Cornell University)

    Repository · 38 publications · 3.9K citations

    21

    h-index

  2. Entropy

    2 publications · 30 citations

    2

    h-index

  3. IEEE Transactions on Medical Imaging

    3 publications · 141 citations

    2

    h-index

  4. IEEE Signal Processing Magazine

    1 publications · 28 citations

    1

    h-index

  5. Journal of Econometrics

    1 publications · 22 citations

    1

    h-index

1–5