Towards Healing the Blindness of Score Matching
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2022
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917097196486656 |
|---|---|
| author | Zhang, Mingtian Key, Oscar Hayes, Peter Barber, David Paige, Brooks Briol, François-Xavier |
| author_facet | Zhang, Mingtian Key, Oscar Hayes, Peter Barber, David Paige, Brooks Briol, François-Xavier |
| contents | Score-based divergences have been widely used in machine learning and statistics applications. Despite their empirical success, a blindness problem has been observed when using these for multi-modal distributions. In this work, we discuss the blindness problem and propose a new family of divergences that can mitigate the blindness problem. We illustrate our proposed divergence in the context of density estimation and report improved performance compared to traditional approaches. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2209_07396 |
| institution | arXiv |
| publishDate | 2022 |
| record_format | arxiv |
| spellingShingle | Towards Healing the Blindness of Score Matching Zhang, Mingtian Key, Oscar Hayes, Peter Barber, David Paige, Brooks Briol, François-Xavier Machine Learning Score-based divergences have been widely used in machine learning and statistics applications. Despite their empirical success, a blindness problem has been observed when using these for multi-modal distributions. In this work, we discuss the blindness problem and propose a new family of divergences that can mitigate the blindness problem. We illustrate our proposed divergence in the context of density estimation and report improved performance compared to traditional approaches. |
| title | Towards Healing the Blindness of Score Matching |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2209.07396 |