score-models
The aim of score-models is to trace the history and evolution of score-matching models for sampling from a data distribution.
References
@article{hyvarinen2005estimation,
title={Estimation of non-normalized statistical models by score matching.},
author={Hyv{\"a}rinen, Aapo and Dayan, Peter},
journal={Journal of Machine Learning Research},
volume={6},
number={4},
year={2005}
}
@inproceedings{song2020sliced,
title={Sliced score matching: A scalable approach to density and score estimation},
author={Song, Yang and Garg, Sahaj and Shi, Jiaxin and Ermon, Stefano},
booktitle={Uncertainty in Artificial Intelligence},
pages={574--584},
year={2020},
organization={PMLR}
}
@article{song2019generative,
title={Generative modeling by estimating gradients of the data distribution},
author={Song, Yang and Ermon, Stefano},
journal={arXiv preprint arXiv:1907.05600},
year={2019}
}
@article{vincent2011connection,
title={A connection between score matching and denoising autoencoders},
author={Vincent, Pascal},
journal={Neural computation},
volume={23},
number={7},
pages={1661--1674},
year={2011},
publisher={MIT Press}
}
Additional https://courses.cs.washington.edu/courses/cse599i/20au/resources/L17_denoising.pdf How to Train Your Energy-Based Models https://arxiv.org/pdf/2101.03288.pdf A Low Rank Approach to Automatic Differentiation https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.161.7201&rep=rep1&type=pdf Sliced Score Matching https://arxiv.org/pdf/1905.07088.pdf https://arxiv.org/pdf/2101.09258.pdf