Learning nonnegative matrix factorizations from compressed data
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917274892369920 |
|---|---|
| author | Chaudhry, Abraar Rebrova, Elizaveta |
| author_facet | Chaudhry, Abraar Rebrova, Elizaveta |
| contents | We propose a flexible and theoretically supported framework for scalable nonnegative matrix factorization. The goal is to find nonnegative low-rank components directly from compressed measurements, accessing the original data only once or twice. We consider compression through randomized sketching methods that can be adapted to the data, or can be oblivious. We formulate optimization problems that only depend on the compressed data, but which can recover a nonnegative factorization which closely approximates the original matrix. The defined problems can be approached with a variety of algorithms, and in particular, we discuss variations of the popular multiplicative updates method for these compressed problems. We demonstrate the success of our approaches empirically and validate their performance in real-world applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_04994 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning nonnegative matrix factorizations from compressed data Chaudhry, Abraar Rebrova, Elizaveta Optimization and Control Machine Learning Numerical Analysis 15A23, 68T10, 68W20, 65F30 We propose a flexible and theoretically supported framework for scalable nonnegative matrix factorization. The goal is to find nonnegative low-rank components directly from compressed measurements, accessing the original data only once or twice. We consider compression through randomized sketching methods that can be adapted to the data, or can be oblivious. We formulate optimization problems that only depend on the compressed data, but which can recover a nonnegative factorization which closely approximates the original matrix. The defined problems can be approached with a variety of algorithms, and in particular, we discuss variations of the popular multiplicative updates method for these compressed problems. We demonstrate the success of our approaches empirically and validate their performance in real-world applications. |
| title | Learning nonnegative matrix factorizations from compressed data |
| topic | Optimization and Control Machine Learning Numerical Analysis 15A23, 68T10, 68W20, 65F30 |
| url | https://arxiv.org/abs/2409.04994 |