A Full Adagrad algorithm with O(Nd) operations
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_ | 1866915174320963584 |
|---|---|
| author | Godichon-Baggioni, Antoine Lu, Wei Portier, Bruno |
| author_facet | Godichon-Baggioni, Antoine Lu, Wei Portier, Bruno |
| contents | A novel approach is given to overcome the computational challenges of the full-matrix Adaptive Gradient algorithm (Full AdaGrad) in stochastic optimization. By developing a recursive method that estimates the inverse of the square root of the covariance of the gradient, alongside a streaming variant for parameter updates, the study offers efficient and practical algorithms for large-scale applications. This innovative strategy significantly reduces the complexity and resource demands typically associated with full-matrix methods, enabling more effective optimization processes. Moreover, the convergence rates of the proposed estimators and their asymptotic efficiency are given. Their effectiveness is demonstrated through numerical studies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_01908 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Full Adagrad algorithm with O(Nd) operations Godichon-Baggioni, Antoine Lu, Wei Portier, Bruno Statistics Theory Machine Learning A novel approach is given to overcome the computational challenges of the full-matrix Adaptive Gradient algorithm (Full AdaGrad) in stochastic optimization. By developing a recursive method that estimates the inverse of the square root of the covariance of the gradient, alongside a streaming variant for parameter updates, the study offers efficient and practical algorithms for large-scale applications. This innovative strategy significantly reduces the complexity and resource demands typically associated with full-matrix methods, enabling more effective optimization processes. Moreover, the convergence rates of the proposed estimators and their asymptotic efficiency are given. Their effectiveness is demonstrated through numerical studies. |
| title | A Full Adagrad algorithm with O(Nd) operations |
| topic | Statistics Theory Machine Learning |
| url | https://arxiv.org/abs/2405.01908 |