Stochastic Weakly Convex Optimization Beyond Lipschitz Continuity
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_ | 1866913572278239232 |
|---|---|
| author | Gao, Wenzhi Deng, Qi |
| author_facet | Gao, Wenzhi Deng, Qi |
| contents | This paper considers stochastic weakly convex optimization without the standard Lipschitz continuity assumption. Based on new adaptive regularization (stepsize) strategies, we show that a wide class of stochastic algorithms, including the stochastic subgradient method, preserve the $\mathcal{O} ( 1 / \sqrt{K})$ convergence rate with constant failure rate. Our analyses rest on rather weak assumptions: the Lipschitz parameter can be either bounded by a general growth function of $\|x\|$ or locally estimated through independent random samples. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_13971 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Stochastic Weakly Convex Optimization Beyond Lipschitz Continuity Gao, Wenzhi Deng, Qi Optimization and Control Machine Learning This paper considers stochastic weakly convex optimization without the standard Lipschitz continuity assumption. Based on new adaptive regularization (stepsize) strategies, we show that a wide class of stochastic algorithms, including the stochastic subgradient method, preserve the $\mathcal{O} ( 1 / \sqrt{K})$ convergence rate with constant failure rate. Our analyses rest on rather weak assumptions: the Lipschitz parameter can be either bounded by a general growth function of $\|x\|$ or locally estimated through independent random samples. |
| title | Stochastic Weakly Convex Optimization Beyond Lipschitz Continuity |
| topic | Optimization and Control Machine Learning |
| url | https://arxiv.org/abs/2401.13971 |